DeOS AlphaBook

Chapter 12 · Part V · The practice: How it is delivered

Consulting workbook

pages 355–45790 figures49 tables

Handbook and workbook · Structure, method, templates

1. The consulting workbook#

From the first engineered decision to a decision system: one working book for clients and the vencortex team#

A maturity assessment with a decision heat map, use case templates for an analysis, a decision policy and a solution, the engineering, value and gate sheets that take a use case from Discovery to Scaling, and the DeOS screens where each template lands.

01 · assess 02 · define 03 · engineer
Six dimensions, five roles Three nested templates Prove with the same sheets
Questionnaire, evidence rules and a heat map set the entry level. Analysis, decision policy and solution, each with a worked example. Audit, causal graph, criteria, scoring, underwriting, gates, roadmap.

Summary

2. In brief#

The Structure Handbook explains what a decision is made of in DeOS. The Delivery and Consulting Framework explains how we take a firm from its first engineered decision to a decision system. This book puts both in one place and turns them into work: every concept comes with the template that captures it, and every template shows the DeOS screen it fills.

One framework, three levels.

Decision optimization is Strategy, Intelligence and Platform. A solution is the decision lifecycle of a domain, a decision policy formalizes one decision and combines its analyses through its own Recommend step, and an analysis runs four components.

Six phases, fast and low risk.

Discovery takes one to two weeks and the Audit one to two weeks per decision, both read only; a dedicated instance is set up in days; the proof of value runs in parallel to the live process and can stop at the gate.

Assess before you choose.

A maturity questionnaire covers six dimensions tied to the framework, with a common core and a module for each of five roles. Stages 3 and 4 count only with evidence. A value-weighted decision heat map turns scores into a recommended entry level.

Define use cases at the right level.

Template A defines an analysis, with a human / AI mode per component; Template B a decision policy with type, objective, owner, constraints, recommendation and analyses; Template C an end-to-end solution. An archetype directory gives each its starting point.

Score at three levels.

Desirability, viability, feasibility and risk are asked differently for a solution, a policy and an analysis, and roll up: reuse raises viability, and a red-line risk below stops everything above it.

Engineer with the sheets that produce the assets.

Decision audit, causal decision graph canvas, goals, options and actions, perspectives and criteria library, insights and expert questions, context and data room, and process, governance and the decision record.

Qualify, underwrite, gate and prove with the same fields.

Long list and D·V·F·R scoring, alpha underwriting, readiness check, gate review, proof plan, roadmap and decision portfolio use the same fields from Discovery to Scaling. Decisions are judged against the probabilities they were made under, not one outcome.

From workbook to DeOS.

Every use case template lands on a DeOS screen: the solution, the decision policy overview, the process, insights and expert assessment, and the recommendation. Template codes match everywhere, and worked examples follow one PE buy-and-build platform.

6maturity dimensions, from decision governance to scale and capability, each scored on five stages and tied to a part of the framework.
5roles the questionnaire has a module for, on top of a common core: sponsor, decision owner, investment professional, control, and data and technology.
1 to 2weeks for Discovery, and one to two weeks per decision for the Audit, both read only, with no commitment beyond the workshop.

The framework

3. The framework: three levels, one map#

Decision optimization is Strategy, Intelligence and Platform. Every use case lives at one of three levels of that framework: a solution, a decision policy or an analysis. The level decides what is built, who owns it, how it is proven, which maturity level it belongs to and what the Scale gate can scale.

  1. STRATEGY

    End-to-end decision policy + mandate Every decision across the decision lifecycle is connected and governed by one mandate

  2. INTELLIGENCE

    Hybrid intelligence: Perceive, Diagnose, Predict, Recommend Every analysis runs from context to the recommended action and its explanation

  3. PLATFORM

    Governance across every level, plus Context, Analysis, Recommendation, Learning Every solution, decision policy and analysis is governed, executed and learned from in one system

Three levels#

Level Definition Contains Owner In DeOS
Solution The decision lifecycle of several decisions for one domain: use-case specific stages, each holding a set of decisions, run by named stakeholders under one mandate. Stages, decision policies, hand-offs, decision assets Solution owner (CIO, head of strategy, managing partner) Studio · Solutions
Decision policy One decision, formalized. It combines its analyses into one recommendation through its own Recommend step. Type, objective, owner, constraints, recommendation, analyses, process Decision owner (PM, deal partner, CFO, committee) Studio · Decision policies
Analysis One run of the intelligence flow, built from four components: Perceive context, Causal diagnosis, Counterfactual prediction, Contextual optimization. Four components, each built from analysis archetypes, with a human / AI mode each Analysis owner (analyst, associate, operating partner) Studio · Analysis builder

Four components of every analysis#

An analysis is one run of the intelligence flow. Its four components are the four steps Perceive, Diagnose, Predict and Recommend. Each is built from analysis archetypes of the catalogue and runs on a platform capability. Diagnose and Predict run on the Analysis capability, not on Context.

Component Step Capability Question Analysis steps of the catalogue
Perceive context Scope Context What should we look at, and what is it comparable to? Context engineering: representation, scan, construction, generation
Causal diagnosis Diagnose Analysis What really drives outcomes here, and where do we stand? Causal diagnosis: valuation, mispricing, driver decomposition, regime, relative performance, rating, anomaly
Counterfactual prediction Predict Analysis What will, could and would happen? Probabilistic prediction and fusion (will); counterfactual simulation (could, would)
Contextual optimization Recommend Recommendation What should we do, given our goals and constraints, and why? Contextual optimization (Selection, policy optimization, Go / No-Go); explanation

Four decision policy types#

Every decision policy has one of four types: Selection, Policy optimization, Go / No-Go or Monitoring. The type sets the question its Recommend step answers and the shape of its output. Inside an analysis, contextual optimization uses Selection, policy optimization or Go / No-Go, always with an explanation; monitoring sets triggers and explains why outcomes differed.

Type Question Output Decision archetypes
Selection Which options? A ranked, scored shortlist Selection
Policy optimization Which actions? The best action or setting: price, lever, timing, structure, weights Price, treatment, timing, financing, allocation, portfolio construction, hedging
Go / No-Go Yes or no? A binary decision with its conviction threshold and the terms under which it holds Conviction test
Monitoring What should trigger action, and why did it differ? Triggers, thresholds and early warnings; attribution of variance to its causes Trigger design, root cause

Five platform capabilities#

CAPABILITY
Governance
WHAT IT DOES
Spans every level: solutions, decision policies and analyses. Codifies policies and the mandate as auditable

processes with guardrails and constraints; keeps the audit trail of which insight was created on what data, by which analysis and by which person; versions policies, analyses and recommendations; holds the decision assets and DeOS Studio.

Context

Builds the context through data connectors to the client's proprietary sources, internal systems and third-party data providers: data rooms, ontology, entities, drivers and comparables.

CAPABILITY WHAT IT DOES CAPABILITY WHAT IT DOES

Analysis

Runs causal diagnosis and counterfactual prediction as one end-to-end analysis: regime-aware, causal and probabilistic foundation models propagate predictions from one model to the next, fused with expert judgment, with the human brought in at defined points.

RecommendationTurns the analyses into a decision recommendation: recommended actions, probabilistic expected outcomes on the optimization goal, conviction, ranked options, aggregation from asset to portfolio, and a fully explainable why chain from several perspectives.

Learning

Compares outcomes with predictions and updates beliefs, models and policies, on the client's instance alone.

STRUCTURE · WHAT IS BUILT USE CASE TEMPLATE MATURITY LEVEL UNIT OF VALUE · SCALE Solution L4 End-to-end decision lifecycle Solution SOLUTION C use-case specific stages, each a set of decisions · stakeholders L5 Total portfolio use case a decision lifecycle or the total portfolio Thesis Platform Add- STRUCTURE · WHAT IS BUILT USE CASE TEMPLATE MATURITY LEVEL UNIT OF VALUE · SCALE GOVERNANCE · ACROSS EVERY LEVEL Solution use-case specific stages, each a set of decisions · stakeholders Thesis Platform Add-ons Value cr. Exit Decision policy one decision; its Recommend step combines its analyses Type Objective Owner Constraints Recommend. Analysis one intelligence flow · four components Perceive context Causal diagnosis Counterfact. prediction Contextual optimization C Solution use case L4 End-to-end decision lifecycle L5 Total portfolio SOLUTION a decision lifecycle or the total portfolio B Decision policy use case L1 Decision governance L3 Single policy DECISION a policy rolled out to teams and funds A Analysis use case L2 Analysis augmented ANALYSIS an analysis reused across decisions
Figure 1 The three levels of the framework (left) are the levels at which use cases are defined (Templates A, B, C), the maturity levels of the Delivery Framework, and the three things the Scale gate scales. Governance spans all three levels: solutions, decision policies and analyses are governed in one system. The maturity level L1 decision governance starts it with decision policies, even when they run manually.
A · analysis B · decision policy C · solution
One analysis: four components, each built from analysis archetypes and with a human / AI mode One decision: type, objective, owner, constraints, recommendation, the analyses it combines, process One decision lifecycle: use-case specific stages, the set of decisions at each stage, stakeholders, one mandate, decision assets
Is this input right, and better than the status quo? Is this decision made better, and does it turn out better? Do the decisions of a strategy work together, without loss at hand-offs?
Analysis owner: analyst, associate, operating partner Decision owner: PM, deal partner, CFO, committee Solution owner: CIO, head of strategy, managing partner
L2 Analysis augmented L1 Decision governance, L3 Single decision policy L4 End-to-end decision lifecycle, L5 Total portfolio approach
Information and analysis alpha All four alphas for one decision Alpha across the decision lifecycle; allocation alpha
Analysis: calibration, accuracy against base rates and human-only Outcome and process: backtest, parallel run, adherence Outcome, process and innovation across decisions
Scale an analysis Scale a decision policy Scale a solution

Structure object

Core question

Typical owner

Maturity level

Alpha it moves first

Kind of proof

Scale outcome

The nesting rule. A solution use case lists its stages and the set of decisions at each; every decision is a decision use case. A decision use case lists its analyses; each is an analysis use case, new or reused. An analysis use case lists every decision it serves. Defined this way, the second use case starts from the first, which is what makes scaling cheap.

02

The process

4. Six phases, their templates, effort and risk#

Every use case moves through the same six phases and four gates. Each phase fills a defined set of templates; each gate reads them. Most of the path is read only or runs in parallel to the live process, so the client can stop at every gate.

1 2 3 4 5 6 Implementation Discovery Audit Engineering Run Scaling / PoV M Maturity D Decision audit C Solution E N Decision record Q Gate review Expert questions H Heat map V B I U S Underwrite Decision policy Insights Evaluation Roadmap L B A Analysis F V C Solution update Long list Policy (draft) 1 Discovery M Maturity H Heat map L Long list V Underwrite* R Readiness 2 Audit D Decision audit V Underwrite B Policy (draft) C Solution (draft) G1 DISCOVER 3 Engineering C Solution B Decision policy A Analysis G CDG canvas O Goals, options K Criteria X Context P Process 4 Implementation / PoV E Expert questions I Insights F Proof plan W Evaluation plan N Decision record G2 PROVE 5 Run N Decision record U Evaluation V Realised Q Gate review G3 REALISATION 6 Scaling Q Gate review S Roadmap C Solution update M Re-assess H Heat map G4 SCALE * first, rough underwriting on the long list
Figure 2 The six phases with the templates each one fills. Gates sit at the end of a stage: G1 Discover after the Audit, G2 Prove after Implementation / PoV, G3 Realisation after Run, G4 Scale after Scaling.
Phase Key question Templates filled Output and gate
1 · Discovery Which decisions are worth engineering? M maturity questionnaire, H decision heat map, L long list scored at three levels, first V underwriting, R readiness Maturity profile, recommended entry level, scored long list, potential and complexity map, roadmap
2 · Audit What does each decision cost today, and what could it earn? D decision audit, V underwriting in full, first cut of B (identity and status quo) and C (stages and decisions) Audit and underwritten alpha per use case · G1 Discover
3 · Engineering How should this decision be made, beyond human limits? C solution, B decision policy, A analyses with human / AI modes, G causal decision graph, O goals and actions, K criteria, X context, P process Solution, policy and analysis templates filled; a policy ready to configure
4 · Implementation / PoV Does it work on a real decision? E expert questions, I insights, F proof plan, W evaluation plan, N records of the first decisions Evidence on own decisions, verified alpha · G2 Prove
5 · Run Is it part of how we work? N decision records, U decision evaluation, V realised against underwritten, Q value-realised review Live use case · G3 Realisation
6 · Scaling What next, and who builds it? Q scale review, S roadmap and decision portfolio, C solution update, M and H re-assessment Next round of the portfolio · G4 Scale

Duration, client effort and risk to the client#

Phase Typical duration Client effort Risk to the client
1 · Discovery 1 to 2 weeks Maturity questionnaire in its investor or enterprise version (about 30 minutes per person), one sponsor workshop None: no data leaves the firm, no commitment beyond the workshop
2 · Audit 1 to 2 weeks per decision Two to three hours with each decision owner; read-only access to past decisions None: read only, on the client's premises or instance
3 · Engineering 1 to 3 weeks Working sessions with the decision owner and experts None: design work on paper and in DeOS Studio, nothing in the live process changes
4 · Implementation / PoV Instance in days; proof of value over a few weeks of real decisions Run one live decision with the vencortex team, or let it run in parallel Low: dedicated instance, parallel run, the live book is untouched; stop at the gate
5 · Run Training within a few hours Day-to-day use by the decision owner and team Controlled: people decide, every override is possible and recorded
6 · Scaling Next use case starts from the decision assets The team learns to build in the no-code DeOS Studio Low: each new use case passes the same gates

Durations are typical for one use case and are set per engagement; they are not guarantees.

Fast, by design#

  1. Days

    Dedicated instance and the first use case set up and tested Connectors to existing sources, archetype templates, no system replacement

  2. Weeks

    First decisions run with the team, or in parallel to the live process One decision, one owner, a measurable outcome

  3. Hours

    Team trained for day-to- day use It works on the decision the team already has to make

  4. Each next

    use case Starts from the solution, policy and analysis assets already built Shared ontology, criteria, graphs, expert panel and analyses

Low risk, by design#

  • SAFEGUARD · WHAT IT MEANS FOR THE CLIENT
  • Parallel run · A use case runs next to the existing process on the same decisions until it has passed the Prove gate. The live book is not touched.
  • Gates with a stop · Four gates, each with a stop or park outcome. Nothing moves on because of momentum.
  • Small first scope · One decision, one owner and a measurable outcome. The alpha is underwritten before any engineering.
  • People decide · Recommendations are explained and every override is possible and recorded. Judgement-only parts stay with people by design.
  • Your data, your instance · A dedicated instance on the cloud of the client's choice or on premises; encrypted in transit and at rest;

proprietary models, nothing sent to third parties. Client data is never used to train shared models; the IP stays with the client.

  • Nothing is lost · If a use case stops, the client keeps the audit, the engineered policy and the decision records.
  • Works alongside · Connects to existing data and systems; nothing has to be replaced to start. Governance can start with no models at all.

Light where it can be. A use case entered as a live, fully serviced decision fills Templates B and N with the client and leaves the rest to the vencortex team. A parallel run adds F. A decision portfolio fills L and Q for every use case in the portfolio. The templates are the same; how much the client writes in them is a choice.

03

Roles

5. Roles and responsibilities#

Decision optimization changes how people decide, so the people who decide are part of every phase. Six roles cover most firms; one person can hold several. The sponsor is a stakeholder role; the four decision roles (decision architect, decision maker, recommender, analyst) sit with the decision architect, the decision owner, who recommends or decides as the policy says, and the investment professional.

SPONSOR DECISION OWNER INVESTMENT PROFESSIONAL CIO, CEO, managing partner PM, deal partner, CFO, head of Analyst, associate, operating asset class partner Owns the programme and the portfolio of use cases; often the solution owner. Signs Owns one decision policy: recommends Owns analyses and expert SPONSOR CIO, CEO, managing partner Owns the programme and the portfolio of use cases; often the solution owner. Signs Gate 1 and Gate 4. Answers module R1 of the questionnaire. DECISION OWNER PM, deal partner, CFO, head of asset class Owns one decision policy: recommends or decides, records conviction and overrides. The decision stays with this role. Answers module R2. INVESTMENT PROFESSIONAL Analyst, associate, operating partner Owns analyses and expert inputs, and sets the human / AI mode of each component with the decision architect. Answers module R3. CONTROL Risk, compliance, legal, IC secretary Owns guardrails, model risk and regulatory classification. Checks readiness and trust at every gate. Answers module R4. DATA AND TECHNOLOGY COO, CTO, head of data Owns data access, integration, security and the dedicated instance. Fills Template X. Answers module R5. DECISION ARCHITECT CIO or head of strategy; vencortex at first Engineers solutions, policies, graphs and analyses in DeOS Studio. vencortex acts as the first decision architect from Discovery to Implementation and hands the role over to the CIO or head of strategy and their staff for Run and Scaling.

Who does what, by phase#

Role Discovery 2 audit Engineering 4 impl. / PoV 5 run 6 scaling
Sponsor A A I C I A
Decision owner C R A A A C
Investment professional C C R R R C
Risk and compliance C C C R C C
Data and technology C C R R I R
Decision architect R R R R C R

1

3

R responsible for the work · A accountable for the result · C consulted · I informed. Gate boards are formed from A and C roles of the stage that ends.

Role names by type of firm#

Firm type Sponsor Decision owner Investment prof. Control Data and tech.
Hedge fund (L/S, credit) CIO Portfolio manager Analyst Risk, compliance COO, CTO
Private equity Managing partner Deal partner Associate, operating partner IC, fund counsel CFO, data lead
Asset owner, TPA CIO Head of asset class Investment analyst Risk and governance committee Operations
Family office Principal, CIO Investment director Analyst Board, external counsel Operations
Corporate CEO, CFO BU head, corporate development lead Strategy, FP&A analyst Internal audit, legal CIO, IT

Part i i

6. Maturity assessment#

A short, structured assessment shows where the decision foundation of a firm stands, dimension by dimension and role by role. It does not decide where to start; value does. It shows what a starting point will need.

04

The maturity model

05

Running the assessment

06

Questionnaire by segment and role

07

Scoring and interpretation

08

Decision heat map and entry level

04

Maturity model

7. The maturity model#

Six dimensions describe the decision foundation of a firm. Each is scored on five stages and tied to a part of the framework, so a low score points to the capability that closes it. The stages describe the foundation; the five maturity levels of the Delivery Framework describe what DeOS runs. The two are linked, but not the same.

Dimension What good looks like Alpha most affected
Decision governance Decisions follow defined policies; records, rationale, overrides and audit trail exist. Governance
Decision structure Goals, options, criteria and causal drivers of each decision are explicit and shared. Governance, analysis
Context and data The data a decision needs is decision-centric, point in time, mapped to business objects, with comparables at hand. Information
Prediction and judgment Forecasts are probabilistic; expert judgment is structured and calibrated; AI is fused in a defined way. Analysis, behavioural
Learning and proof Outcomes are compared with expectations; calibration, overrides and root causes feed back into the policy. Behavioural, governance
Scale and capability Assets are reused across teams and funds; the team can build and change decisions itself. All four, across the firm

d1

d2

d3

d4

d5

d6

Five stages per dimension#

5 · Systemic Connected across the decision lifecycle and the firm; improves from every outcome.

1 · Ad hoc

2 · Documented

3 · Structured

4 · Engineered

Depends on individuals;

Written down and partly

Explicit and standard for the

Built into the workflow,

nothing is repeatable or

shared, but not enforced or

key decisions; measured in

probabilistic, calibrated and

recorded.

measured.

parts.

explained.

Tied to the framework, assessed at its level#

Each dimension belongs to a part of Strategy, Intelligence and Platform and is assessed at the level where it lives: governance at the firm and policy, context at the analysis, scale at the solution and firm.

Dimension Framework Assessed at A low score points to
D1 Decision governance Platform · Governance Solution, policy and analysis Governance capability: policies, guardrails, decision records
D2 Decision structure Strategy · policy and mandate Policy and solution Template B: objective, criteria, causal decision graph
D3 Context and data Intelligence · Perceive / Platform · Context Analysis Perceive context of the key analyses; Template X
D4 Prediction and judgment Intelligence · Diagnose, Predict, Recommend / Platform · Analysis, Recommendation Analysis and policy Causal diagnosis, counterfactual prediction and expert fusion; human / AI modes
D5 Learning and proof Platform · Learning Policy and solution Learning capability: outcomes, calibration, root cause
D6 Scale and capability Solution · decision assets Solution and firm Decision assets of a solution; the team building in DeOS Studio

Stage descriptors#

1 ad hoc 2 documented 3 structured 4 engineered 5 systemic
Decisions depend on who is in the room; little is recorded Process documented; IC minutes, but conviction and alternatives not recorded Decision policies for key decisions; conviction, rationale and overrides recorded Policies enforced in the workflow; guardrails block incomplete decisions; full audit trail One governance model across all strategies and funds; records used for proof of skill
Goals implicit; criteria differ by person Templates and checklists; goals stated but not measurable Measurable goals, shared criteria with thresholds and weights Causal decision graph per decision type; options and actions analysed in full Decisions connected along the decision lifecycle; decision lifecycles optimized; one decision language firm-wide
Data collected by hand per decision Central data, organized by source system; peers chosen by habit Data room per decision type; comparables scanned systematically Ontology of business objects; point in time; context scan automated Proprietary data foundation reused by every decision; context generated where data is scarce
Single-point forecasts and opinions Ranges for some inputs; models used ad hoc Probabilistic forecasts for key analyses; expert inputs structured Experts weighted by calibration; AI functions fused with experts; human / AI mode set per component; every prediction explained Chained analyses with uncertainty carried through to portfolio and total-portfolio decisions
Outcomes not compared with expectations Post-mortems on large losses only Outcomes linked to decisions; overrides tracked Calibration feedback to each expert; root cause runs along the causal graph Policies, weights and graphs updated systematically from outcomes across the firm
Every analysis built from scratch Some templates shared within one team Analyses and criteria reused across decisions of one strategy Policies rolled out across teams and funds; team builds in the no-code DeOS Studio Firm-wide decision system; new strategies launched on existing assets

D1 Governance

D2 Structure

D3 Context

D4 Prediction

D5 Learning

D6 Scale

Evidence rule per stage#

A stage counts only when its evidence can be shown. Stages 1 and 2 are rated; stages 3 and 4 need the artefacts below; stage 5 needs stage 4 evidence across more than one strategy. The score of a dimension is the lower of the rating and the evidence.

Dimension Evidence required for stage 3 Evidence required for stage 4
D1 Decision governance Written policies for the key decisions; three past decisions reconstructable Guardrails in the workflow; full decision records with overrides and reasons
D2 Decision structure Measurable goals and shared criteria with thresholds for key decisions A causal decision graph per decision type; options analysed in full
D3 Context and data A data room per decision type; comparables scanned systematically An ontology of business objects; point-in-time data; automated scan
D4 Prediction and judgment Probabilistic forecasts for key analyses; expert input structured Experts weighted by calibration; human / AI mode set per component
D5 Learning and proof Outcomes linked to decisions; overrides tracked Calibration feedback per expert; root cause along the graph
D6 Scale and capability Analyses and criteria reused across decisions of one strategy Policies rolled out across teams; the team builds in DeOS Studio

Stages are not entry levels. A firm at stage 2 on context can still enter at L3 with one decision policy: the decision-centric data room integrates only what that decision needs. Low stages raise complexity and readiness work; they do not rule a use case out. Chapter 07 shows how the scores feed the D · V · F · R scoring; chapter 08 turns them, decision by decision, into a recommended entry level.

05

Assessment

8. Running the assessment#

The assessment takes little time from each person and produces the first evidence of the engagement: where the foundation is, where people disagree about it, and which decisions they would engineer first.

1 2 3 4 Select respondents Questionnaire Interviews and evidence Scoring workshop › › › At least one person per role, ideally Common core plus one role Short follow-ups where roles Maturity profile, perception gaps, three decision owners and three module. Rated statements and diverge; the artefacts 1 Select respondents At least one person per role, ideally three decision owners and three analysts per strategy in scope. › 2 Questionnaire Common core plus one role module. Rated statements and three open questions. About 30 minutes, done individually. › 3 Interviews and evidence Short follow-ups where roles diverge; the artefacts for stage 3 and 4; a decision heat map for each decision named. › 4 Scoring workshop Maturity profile, perception gaps, value-weighted heat map and the recommended entry level, discussed with the sponsor.

Rating scale#

Every statement is rated for the firm as it works today, not as it should work. Respondents rate what they see in their own work.

1
Not true. We do not do this.
2
Rarely true. Some individuals,
some of the time.
3
Partly true. For key decisions,
not consistently.
4
Mostly true. Standard practice,
with exceptions.
5
Fully true. Systematic, enforced
and measured.

? marks "I don't know". It is not scored, and it is a finding: if the people who decide do not know whether outcomes are tracked, they are not.

What each part covers#

PART

Two versions

Common core

R1 · E1 Sponsor

R2 · E2 Decision owner

R3 · E3 Professional

R4 · E4 Control

ITEMS

I · E

10 + 3

10 + 3

10 + 3

10 + 3

PURPOSE

Version I for investors, Version E for CEOs and enterprises: the same dimensions, codes and modules, worded for each segment (chapter 06).

Two statements per dimension, answered by every role. Makes roles comparable and reveals perception gaps.

Strategy, value, proof of skill, appetite for AI, programme ownership.

How decisions are actually made: sizing, options, overrides, biases, monitoring.

Time spent, context, forecasting practice, calibration, reuse of work.

Guardrails, audit trail, model risk, AI regulation, trust and user control.

PART ITEMS PURPOSE 10 + 3 R5 · E5 Data and Ontology, point-in-time data, integration speed, deployment, security. PART ITEMS PURPOSE R5 · E5 Data and 10 + 3 Ontology, point-in-time data, integration speed, deployment, security.

technology

H Decision heat map
1 per decision
How each component of a candidate decision is done today, from not done to
automated (chapter 08).

What makes this assessment different#

Mapped to the framework

Each dimension is tied to a part of Strategy, Intelligence and Platform, so a low score points to the capability that closes it.

Human / AI today

For each candidate decision, the heat map records how each component is done today: not done, judgement only, augmented or automated.

The open questions are the start of discovery. Every module asks the respondent to name the decisions that matter most to them, how often they are made and what is at stake. These answers become the rows of the decision heat map (Template H) and the first rows of the long list (Template L), scored with the same D · V · F · R grid.

Evidence, not opinion

A stage counts only when its evidence can be shown. Stages 3 and 4 need the artefacts listed per dimension; the score is the lower of rating and evidence.

Weighted by value

Gap × value at stake ranks where the gap matters. A large gap on a small decision is not a priority.

Three levels

Firm scores come from the questionnaire. Each candidate decision gets a decision heat map; each candidate solution a hand-off check across its stages.

Entry, not a ladder

The result is a recommended entry level (L1 to L5), the audit scope for phase 2 and the readiness work. It is repeated at Scaling to show progress.

06

Questionnaire

9. Questionnaire by segment and role#

The questionnaire comes in two versions with the same six dimensions, the same twelve-item core and the same five role modules: Version I for investors (asset managers, hedge funds, private equity, credit, family offices) and Version E for CEOs and enterprises. Only the wording and the examples differ, so scores are comparable across segments. Every respondent answers the common core and the module for their role, rates each statement 1 to 5 for the firm as it works today, and marks ? if they do not know.

Which version, and how the words map#

Use Version I when the firm's decisions are investments: positions, deals, allocations. Use Version E when they are business decisions: capital allocation, pricing, M&A, product, operations, hiring. A corporate with an investment arm takes both, one per population. The dimensions, codes and scoring are identical; the vocabulary maps as follows.

  • VERSION I · INVESTORS · VERSION E · CEOS AND ENTERPRISES
  • Alpha; value above the benchmark · Value created above plan; margin, cash and growth above the market
  • PM, deal partner, head of asset class · Business unit head, CFO, COO, CCO, head of product
  • Position, deal, name · Initiative, commitment, project, market
  • Investment committee · Executive committee, board, steering committee
  • Fund, strategy, sleeve · Business unit, market, product line
  • Thesis; entry, add, trim, exit · Business case; start, scale, pause, stop
  • Mandate and investment policy · Strategy and delegation of authority
  • Analyst, associate, operating partner · Strategy, FP&A, product and pricing manager, M&A team

Same scores, different words. A statement keeps its code (C1 to C12, module item 1 to 10) and its dimension tag in both versions, so Scoring (chapter 07), the heat map and the maturity profile work unchanged, and a group's investor and enterprise populations can be read side by side.

Questionnaire · Version i · Investors

10. Version I · Investors#

For asset managers, hedge funds, private equity and credit funds, family offices and allocators. Modules R1 to R5 follow the common core.

Respondent Name, role, team or fund Date and scope Strategies or funds you answer for Respondent Name, role, team or fund Date and scope Strategies or funds you answer for

Common core · all roles#

1 2 3 4 5 1 2 3 4 5
# statement dim. ?
C1 Our important investment decisions follow a documented process that is the same whoever is in the room. D1
C2 For a decision made a year ago, we can reconstruct what was known, the alternatives considered, who decided and why. D1
C3 Before a decision, we state its goal and target outcome in measurable terms. D2
C4 We judge cases with explicit, shared criteria that have defined weights or thresholds. D2
C5 The data a decision needs is available in one place, point in time, without manual collection. D3
C6 We compare each case systematically with truly comparable assets and past situations. D3
C7 Judgments are given as probabilities or ranges, not as single numbers or opinions. D4
C8 Model or AI outputs feed our decisions in a defined and transparent way. D4
C9 We compare decision outcomes with what we expected and feed the lessons back into how we decide. D5
C10 We know how well calibrated the forecasts of our analysts and PMs are. D5
C11 Methods, models and templates built for one team or fund are reused by others. D6
C12 Our team can define or change a decision process or analysis without an IT project. D6
D1 governance · D2 structure · D3 context and data · D4 prediction and judgment · D5 learning and proof · D6 scale and capability.

Questionnaire · Version i · Module r1

11. Sponsor#

For: CIO, CEO, managing partner. Answer after the common core.

# STATEMENT DIM. 1 2 3 4 5 ? D2 R1.1 I can name the five decisions that drive most of our alpha, and how often each is made. D1 R1.2 Investment committee decisions record conviction, rationale and overrides. D5 R1.3 We can separate skill from luck in the track record of each PM and strategy. D6 R1.4 # STATEMENT DIM. 1 2 3 4 5 ? R1.1 I can name the five decisions that drive most of our alpha, and how often each is made. D2 R1.2 Investment committee decisions record conviction, rationale and overrides. D1 R1.3 We can separate skill from luck in the track record of each PM and strategy. D5 R1.4 Strategies and funds share one decision language: goals, criteria, risk. D6 R1.5 Allocation across strategies, funds and sleeves follows an explicit total-portfolio framework. D2 R1.6 We measure the value our process adds, not only the returns of the portfolio. D5 R1.7 We have a clear stance on where AI should and should not inform decisions. D4 R1.8 There is a sponsor and a budget for improving how decisions are made, distinct from data and IT projects. D6 R1.9 When investors or regulators ask why we made a decision, we give a complete answer quickly. D1 R1.10 Proprietary data and knowledge stay with the firm when people leave. D3

Open questions#

Which three decisions would you engineer first, how often are they made, and what is at stake in each? What would make you call this programme a success in twelve months? What risk would make you stop it? Which three decisions would you engineer first, how often are they made, and what is at stake in each? What would make you call this programme a success in twelve months? What risk would make you stop it?

Questionnaire · Version i · Module r2

12. Decision owner#

For: PM, deal partner, head of asset class, CFO. Answer after the common core.

# STATEMENT DIM. 1 2 3 4 5 ? D2 R2.1 My position sizes follow from conviction in a defined, repeatable way. D2 R2.2 For each position, entry, add, trim and exit conditions are written down. D4 R2.3 I consider several options (size, timing, instrument, structure) before deciding, not only yes or no. # STATEMENT DIM. 1 2 3 4 5 ? R2.1 My position sizes follow from conviction in a defined, repeatable way. D2 R2.2 For each position, entry, add, trim and exit conditions are written down. D2 R2.3 I consider several options (size, timing, instrument, structure) before deciding, not only yes or no. D4 R2.4 I see the drivers of a thesis and what the market is pricing, not only the forecast. D3 R2.5 When a position goes wrong, we decompose the loss into market, factor, thesis and timing. D5 R2.6 Analyst inputs reach me in a consistent format that I can compare across names and deals. D4 R2.7 When I override a model, a rule or a recommendation, I record why, and the override is reviewed later. D1 R2.8 I get feedback on my own decision biases, such as holding losers too long. D5 R2.9 Thesis-break signals are monitored without my manual attention. D3 R2.10 I can cover more names or deals without lowering decision quality. D6

Open questions#

Which decision takes most of your time, and how often do you make it? On which decision would you most value a second, independent opinion? Which recurring mistake would you most like to remove from your process? Which decision takes most of your time, and how often do you make it? On which decision would you most value a second, independent opinion? Which recurring mistake would you most like to remove from your process?

Questionnaire · Version i · Module r3

13. Investment professional#

For: Analyst, associate, operating partner. Answer after the common core.

# STATEMENT DIM. 1 2 3 4 5 ? D3 R3.1 Most of my time goes into analysis and judgment, not into collecting and cleaning data. D3 R3.2 I can find comparable companies and past set-ups for a name or deal in minutes. D4 R3.3 I give forecasts as ranges with a stated confidence. D5 R3.4 I know how accurat # STATEMENT DIM. 1 2 3 4 5 ? R3.1 Most of my time goes into analysis and judgment, not into collecting and cleaning data. D3 R3.2 I can find comparable companies and past set-ups for a name or deal in minutes. D3 R3.3 I give forecasts as ranges with a stated confidence. D4 R3.4 I know how accurate my past forecasts were, by type of question. D5 R3.5 I know which criteria the PM or committee will use to judge my recommendation. D2 R3.6 The models and AI tools I use explain their outputs in terms I can check. D4 R3.7 I can state the causal drivers of my thesis and how much each moves the outcome. D2 R3.8 My models and notes are stored so that others can find and reuse them. D6 R3.9 Dissenting views in the team are captured, not lost in the discussion. D4 R3.10 The inputs required before a recommendation are clear and enforced. D1

Open questions#

Which task would you hand to a machine tomorrow? Which judgment would you never hand over, and why? Where do you most lack data or context today? Which task would you hand to a machine tomorrow? Which judgment would you never hand over, and why? Where do you most lack data or context today?

Questionnaire · Version i · Module r4

14. Control#

For: Risk, compliance, legal, IC secretary. Answer after the common core.

# STATEMENT DIM. 1 2 3 4 5 ? D1 R4.1 Mandate limits, investment policy and guardrails are encoded and checked before execution. D1 R4.2 Every decision has a complete, reproducible audit trail. D2 R4.3 Risk criteria are part of the decision itself, not a separate check afterwards. D4 R4.4 Models used # STATEMENT DIM. 1 2 3 4 5 ? R4.1 Mandate limits, investment policy and guardrails are encoded and checked before execution. D1 R4.2 Every decision has a complete, reproducible audit trail. D1 R4.3 Risk criteria are part of the decision itself, not a separate check afterwards. D2 R4.4 Models used in decisions are validated, monitored and documented. D4 R4.5 AI use cases are classified against applicable regulation, including AI regulation. D1 R4.6 Overrides of limits, models or recommendations are tracked with their outcomes. D5 R4.7 Confidential and client data used in decisions is protected and access controlled. D3 R4.8 Users trust model outputs as much as their track record warrants: no blind reliance, no blanket rejection. D4 R4.9 Scenario and downside analysis is standard for material decisions. D2 R4.10 New decision processes pass a defined approval path before they go live. D6

Open questions#

What would a regulator or investor ask about an AI-supported decision? Which controls are manual today and would benefit from being encoded? What is a red-line risk that must stop a use case, however attractive? What would a regulator or investor ask about an AI-supported decision? Which controls are manual today and would benefit from being encoded? What is a red-line risk that must stop a use case, however attractive?

Questionnaire · Version i · Module r5

15. Data and technology#

For: COO, CTO, head of data. Answer after the common core.

# STATEMENT DIM. 1 2 3 4 5 ? D3 R5.1 Data is organized around business objects (companies, deals, assets), not only around source systems. D3 R5.2 Internal and external, structured and unstructured data can be combined for one decision. D3 R5.3 Data is stored point in time, so backtests are free of # STATEMENT DIM. 1 2 3 4 5 ? R5.1 Data is organized around business objects (companies, deals, assets), not only around source systems. D3 R5.2 Internal and external, structured and unstructured data can be combined for one decision. D3 R5.3 Data is stored point in time, so backtests are free of look-ahead. D3 R5.4 A new data source can be integrated in days rather than months. D6 R5.5 Models can be deployed into the decision workflow, not only into reports. D4 R5.6 Decision outcomes are captured as data and linked to the inputs of the decision. D5 R5.7 Integrations with order, portfolio, CRM or deal systems are documented and available. D6 R5.8 Security, tenancy and data-protection requirements for a dedicated instance are defined. D1 R5.9 Data quality, freshness and coverage are monitored. D3 R5.10 Business users can configure workflows and analyses with no-code tools. D6

Open questions#

Which systems hold the data that the top decisions need? Which integration or security constraints should we know about now? Which reviews must a dedicated instance pass before go-live? Which systems hold the data that the top decisions need? Which integration or security constraints should we know about now? Which reviews must a dedicated instance pass before go-live?

Questionnaire · Version e · CEOs and enterprises

16. Version E · CEOs and enterprises#

For corporates and their executive teams: the same six dimensions and codes, worded for business decisions. Modules E1 to E5 follow the common core.

Respondent Name, role, team or fund Date and scope Strategies or funds you answer for Respondent Name, role, team or fund Date and scope Strategies or funds you answer for

Common core · all roles#

1 2 3 4 5 1 2 3 4 5
# statement dim. ?
C1 Our important business decisions (capital allocation, pricing, M&A, product, hiring) follow a documented process that is the same whoever is in the room. D1
C2 For a decision made a year ago, we can reconstruct what was known, the alternatives considered, who decided and why. D1
C3 Before a decision, we state its goal and target outcome in measurable terms: revenue, margin, cash, risk. D2
C4 We judge business cases with explicit, shared criteria that have defined weights or thresholds. D2
C5 The data a decision needs is available in one place, point in time, without manual collection from the business units. D3
C6 We compare each case systematically with comparable past initiatives, markets and competitor moves. D3
C7 Forecasts in business cases are given as ranges or probabilities, not as single numbers. D4
C8 Model, analytics or AI outputs feed our decisions in a defined and transparent way. D4
C9 We compare the outcomes of initiatives with their business case and feed the lessons back into how we decide. D5
C10 We know how well calibrated the forecasts of our planners and business unit heads are. D5
C11 Methods, models and templates built for one business unit are reused by others. D6
C12 Our teams can define or change a decision process or analysis without an IT project. D6
D1 governance · D2 structure · D3 context and data · D4 prediction and judgment · D5 learning and proof · D6 scale and capability.

Questionnaire · Version e · Module e1

17. Sponsor#

For: CEO, board, executive committee. Answer after the common core.

# STATEMENT DIM. 1 2 3 4 5 ? D2 E1.1 I can name the five decisions that create or destroy most of our enterprise value, and how often each is made. D1 E1.2 Executive committee decisions record conviction, rationale and overrides. D5 E1.3 We can separate the effect of our decisions from market condit # STATEMENT DIM. 1 2 3 4 5 ? E1.1 I can name the five decisions that create or destroy most of our enterprise value, and how often each is made. D2 E1.2 Executive committee decisions record conviction, rationale and overrides. D1 E1.3 We can separate the effect of our decisions from market conditions in the results of each business unit. D5 E1.4 Business units share one decision language: goals, criteria, risk. D6 E1.5 Capital and people are allocated across business units and initiatives by an explicit, portfolio-wide framework, not by last year's budget. D2 E1.6 We measure the value our decision process adds, not only the results of the company. D5 E1.7 We have a clear stance on where AI should and should not inform decisions. D4 E1.8 There is a sponsor and a budget for improving how decisions are made, distinct from data and IT projects. D6 E1.9 When the board, investors or regulators ask why we made a decision, we give a complete answer quickly. D1 E1.10 Proprietary data and know-how stay with the company when people leave. D3

Open questions#

Which three decisions would you engineer first, how often are they made, and what is at stake in each? What would make you call this programme a success in twelve months? What risk would make you stop it? Which three decisions would you engineer first, how often are they made, and what is at stake in each? What would make you call this programme a success in twelve months? What risk would make you stop it?

Questionnaire · Version e · Module e2

18. Decision owner#

For: Business unit head, CFO, COO, CCO, head of product. Answer after the common core.

# STATEMENT DIM. 1 2 3 4 5 ? The size and timing of our commitments (budgets, capacity, prices, launches) follow from conviction in a defined, D2 E2.1 repeatable way. D2 E2.2 For each major initiative, the conditions to scale, pause or stop it are written down before it starts. D4 E2.3 I consider se # STATEMENT DIM. 1 2 3 4 5 ? E2.1 The size and timing of our commitments (budgets, capacity, prices, launches) follow from conviction in a defined, repeatable way. D2 E2.2 For each major initiative, the conditions to scale, pause or stop it are written down before it starts. D2 E2.3 I consider several options (scope, timing, partner, structure) before deciding, not only approve or reject. D4 E2.4 I see the drivers behind a business case and what the market and competitors are doing, not only the forecast. D3 E2.5 When an initiative misses its case, we decompose the gap into market, execution, assumption and timing. D5 E2.6 Proposals reach me in a consistent format that I can compare across units and projects. D4 E2.7 When I override a plan, a rule or a recommendation, I record why, and the override is reviewed later. D1 E2.8 I get feedback on my own decision biases, such as escalating commitment to a failing project. D5 E2.9 Early-warning signals on my key initiatives are monitored without my manual attention. D3 E2.10 I can run more initiatives or markets without lowering decision quality. D6

Open questions#

Which decision takes most of your time, and how often do you make it? On which decision would you most value a second, independent opinion? Which recurring mistake would you most like to remove from your process? Which decision takes most of your time, and how often do you make it? On which decision would you most value a second, independent opinion? Which recurring mistake would you most like to remove from your process?

Questionnaire · Version e · Module e3

19. Business professional#

For: Strategy, FP&A, product and pricing managers, M&A team. Answer after the common core.

# STATEMENT DIM. 1 2 3 4 5 ? D3 E3.1 Most of my time goes into analysis and judgement, not into collecting and reconciling data. D3 E3.2 I can find comparable initiatives, markets and competitor moves for a case in minutes. D4 E3.3 I give forecasts as ranges with a stated confidence. D5 E3.4 I know # STATEMENT DIM. 1 2 3 4 5 ? E3.1 Most of my time goes into analysis and judgement, not into collecting and reconciling data. D3 E3.2 I can find comparable initiatives, markets and competitor moves for a case in minutes. D3 E3.3 I give forecasts as ranges with a stated confidence. D4 E3.4 I know how accurate my past forecasts were, by type of question. D5 E3.5 I know which criteria the executive committee will use to judge my proposal. D2 E3.6 The models and AI tools I use explain their outputs in terms I can check. D4 E3.7 I can state the causal drivers of a business case and how much each moves the outcome. D2 E3.8 My models and analyses are stored so that others can find and reuse them. D6 E3.9 Dissenting views in the team are captured, not lost in the discussion. D4 E3.10 The inputs required before a proposal goes to the committee are clear and enforced. D1

Open questions#

Which task would you hand to a machine tomorrow? Which judgement would you never hand over, and why? Where do you most lack data or context today? Which task would you hand to a machine tomorrow? Which judgement would you never hand over, and why? Where do you most lack data or context today?

Questionnaire · Version e · Module e4

20. Control#

For: Risk, compliance, internal audit, legal. Answer after the common core.

# STATEMENT DIM. 1 2 3 4 5 ? D1 E4.1 Delegation of authority, policies and guardrails are encoded and checked before a commitment is made. D1 E4.2 Every material decision has a complete, reproducible audit trail. D2 E4.3 Risk criteria are part of the decision itself, not a separate check afterwards. # STATEMENT DIM. 1 2 3 4 5 ? E4.1 Delegation of authority, policies and guardrails are encoded and checked before a commitment is made. D1 E4.2 Every material decision has a complete, reproducible audit trail. D1 E4.3 Risk criteria are part of the decision itself, not a separate check afterwards. D2 E4.4 Models used in decisions are validated, monitored and documented. D4 E4.5 AI use cases are classified against applicable regulation, including AI regulation. D1 E4.6 Overrides of limits, models or recommendations are tracked with their outcomes. D5 E4.7 Confidential, customer and employee data used in decisions is protected and access controlled. D3 E4.8 Users trust model outputs as much as their track record warrants: no blind reliance, no blanket rejection. D4 E4.9 Scenario and downside analysis is standard for material decisions. D2 E4.10 New decision processes pass a defined approval path before they go live. D6

Open questions#

What would an auditor, a regulator or the board ask about an AI-supported decision? Which controls are manual today and would benefit from being encoded? What is a red-line risk that must stop a use case, however attractive? What would an auditor, a regulator or the board ask about an AI-supported decision? Which controls are manual today and would benefit from being encoded? What is a red-line risk that must stop a use case, however attractive?

Questionnaire · Version e · Module e5

21. Data and technology#

For: CIO, CDO, CTO, head of analytics. Answer after the common core.

# STATEMENT DIM. 1 2 3 4 5 ? D3 E5.1 Data is organized around business objects (customers, products, sites, suppliers), not only around source systems. D3 E5.2 Internal and external, structured and unstructured data can be combined for one decision. D3 E5.3 Data is stored point in time, so past deci # STATEMENT DIM. 1 2 3 4 5 ? E5.1 Data is organized around business objects (customers, products, sites, suppliers), not only around source systems. D3 E5.2 Internal and external, structured and unstructured data can be combined for one decision. D3 E5.3 Data is stored point in time, so past decisions can be evaluated on what was known then. D3 E5.4 A new data source can be integrated in days rather than months. D6 E5.5 Models can be deployed into the decision workflow (planning, pricing, approvals), not only into reports. D4 E5.6 Decision outcomes are captured as data and linked to the inputs of the decision. D5 E5.7 Integrations with ERP, CRM, planning and workflow systems are documented and available. D6 E5.8 Security, tenancy and data-protection requirements for a dedicated instance are defined. D1 E5.9 Data quality, freshness and coverage are monitored. D3 E5.10 Business users can configure workflows and analyses with no-code tools. D6

Open questions#

Which systems hold the data that the top decisions need? Which integration or security constraints should we know about now? Which reviews must a dedicated instance pass before go-live? Which systems hold the data that the top decisions need? Which integration or security constraints should we know about now? Which reviews must a dedicated instance pass before go-live?

07

Scoring

22. Scoring and interpretation#

Scores are simple averages, kept transparent on purpose. The value is in the comparison: across dimensions, across roles, and against the decisions people named.

How scores are computed#

Role score per dimension. The average of all rated statements of that role tagged with the dimension, from the common core and the role module. "?" answers are excluded and counted separately.

Firm score per dimension. The average of the role scores, each role weighted equally, so that a large analyst team does not outvote the sponsor or control.

Evidence cap. A firm score of 3 or more counts only with the stage 3 evidence of chapter 04; 4 or more only with stage 4 evidence. Without it the score is capped at the stage the evidence supports.

Perception gap. The highest minus the lowest role score in a dimension. A gap of 1.5 or more is a finding and is discussed in the workshop.

Unknown rate. The share of "?" answers per dimension. Above 25% the dimension is treated as at most stage 2.

D1 Governance D2 Structure D3 Context D4 Prediction D5 Learning D6 Scale Role avg 3.9 3.4 2.8 2.6 2.4 2.9 3.0 Sponsor 3.1 2.7 2.5 2.4 2.0 2.6 2.6 Decision owner 2.2 2.1 1.8 2.3 1.6 2.2 2.0 Investment professional 3.0 2.6 2.9 2.4 2.2 2.4 2.6 Control 2.8 2.4 2.2 2.6 1.9 2.0 2.3 Data and technology 3.0 D1 Governance D2 Structure D3 Context D4 Prediction D5 Learning D6 Scale Role avg Sponsor 3.9 3.4 2.8 2.6 2.4 2.9 3.0 Decision owner 3.1 2.7 2.5 2.4 2.0 2.6 2.6 Investment professional 2.2 2.1 1.8 2.3 1.6 2.2 2.0 Control 3.0 2.6 2.9 2.4 2.2 2.4 2.6 Data and technology 2.8 2.4 2.2 2.6 1.9 2.0 2.3 Firm score 3.0 2.6 2.4 2.5 2.0 2.4 2.5 Perception gap 1.7 ▲ 1.3 1.1 0.3 0.8 0.9
Figure 3 Illustrative scoring result for a private equity firm running a buy-and-build platform. The sponsor sees governance at 3.9, the investment professionals at 2.2: a perception gap of 1.7. Learning and proof is the weakest dimension for every role. Numbers are illustrative.

Reading the example. Governance looks solid from the top and weak from the deal team: committee minutes exist, but the inputs associates are asked for are neither clear nor enforced. Learning and proof is low everywhere, so effectiveness alpha cannot be shown yet. The first use cases should therefore make decisions recordable and measurable (L1 governance for the investment committee, run together with one L2 analysis such as synergy capture), which raises D1 and D5 at once.

From scores to the engagement#

  • SCORE PATTERN · WHAT IT SUGGESTS
  • D1 below 2.5 · Start with L1 decision governance for the decision with most value, run manually if needed. Every later proof depends on decision records.
  • D1 and D2 at 3 or above, D3 or D4 below 2.5 · Enter at L2: augment the analysis that matters most, with perceive context and calibrated counterfactual prediction, inside the existing process.
  • D1, D2 at 3 or above, D3 at 2.5 or above · Enter at L3: one decision policy end to end, such as add-on go / no-go or credit approval.

or above

  • Several L3 policies live, D6 at 3 or above · Move to L4: connect the stages and decisions of one solution and optimize the hand-offs and decision lifecycles.
  • L4 in two or more strategies, D1 · L5 total portfolio approach: one decision system across strategies, funds and mandates.

and D6 at 3.5 or above

Perception gap of 1.5 or moreTreat as a desirability and adoption risk. Name a champion in the lower-scoring role and include them in the Audit.

Scores also enter the qualification of each use case: low D3 and D4 reduce feasibility, a low D1 on control raises risk, and low D6 lengthens the path to scale. Value is scored separately, from the decision itself.

Workbook · Template

Maturity scoring sheet M PHASE 1 Discovery, again at 6 Scaling COMPLETED BY vencortex team with the sponsor GATE Input to G1 WORKBOOK · TEMPLATE M Maturity scoring sheet PHASE 1 Discovery, again at 6 Scaling COMPLETED BY vencortex team with the sponsor GATE Input to G1

Enter role scores per dimension, then the firm score, the evidence cap, the gap and the unknown rate. Re-run at Scaling to show how the foundation moved.

Role D1 D2 D3 D4 D5 D6 Respondents
Sponsor
Decision owner
Investment professional
Control
Data and technology
Firm score (equal weights)
Evidence shown (stage 3 · 4)
Perception gap (max − min)
Unknown rate
Stage (1 to 5)

Findings#

FINDING
EVIDENCE (SCORES, ARTEFACTS, QUOTES)
ROADMAP

Consequence for the

Decisions named in the open questions#

DECISION
NAMED BY
FREQUENCY
WHAT IS AT STAKE WHAT IS AT STAKE
DECISION NAMED BY FREQUENCY WHAT IS AT STAKE DECISION NAMED BY FREQUENCY WHAT IS AT STAKE

Every decision named here gets a row in the decision heat map (Template H) and on the long list (Template L).

08

Decision heat map

23. Decision heat map and entry level#

The questionnaire scores the firm. The decision heat map scores each candidate decision: how each part of it is done today, from not done to automated. Weighted by the value at stake, it shows where a gap matters, and it turns the assessment into a recommended entry level, the scope of the Audit and the readiness work.

One row per candidate decision, one column per part of the framework#

The columns are the Governance capability, the four components of an analysis (Perceive context, Causal diagnosis, Counterfactual prediction, Contextual optimization) and the Learning capability. Each cell records how that part is done today for this decision, on a five-step scale. The same scale, read forward, is the human / AI mode the engineered decision will use (chapter 11).

0
Not done
J
Judgement only, unstructured
S
Judgement, structured and
recorded
A
Augmented
M
Automated

Value-weighted gap#

Target mode. For each column, the mode the engineered decision should reach. Not every cell should be automated: a final go / no-go stays judgement only, structured and recorded.

Gap. The steps from today to the target, summed over the six columns (0 = not done, J = 1, S = 2, A = 3, M = 4; 0 to 24).

Gap × value. The gap multiplied by the value at stake per year. A large gap on a small decision is not a priority.

Worked example · PE buy and build#

PERCEIVE CAUSAL COUNTERFACTUAL CONTEXTUAL VALUE AT GAP × CANDIDATE DECISION GOVERNANCE CONTEXT DIAGNOSIS PREDICTION OPTIMIZATION LEARNING STAKE GAP VALUE CANDIDATE DECISION GOVERNANCE PERCEIVE CONTEXT CAUSAL DIAGNOSIS COUNTERFACTUAL PREDICTION CONTEXTUAL OPTIMIZATION LEARNING VALUE AT STAKE GAP GAP × VALUE
0 to Target mode A M A A A A $m a year 24 5.0 13 0 65 Add-on go / no-go S J J J J 3.0 11 0 33 Add-on price S A J J J Target mode A M A A A A $m a year 0 to 24 Add-on go / no-go S J J J J 0 5.0 13 65 Add-on price S A J J J 0 3.0 11 33
4.0 11 44 J J A J J J 2.0 15 0 0 30 J J J J J J A J J J 4.0 11 44 J J J 0 J 0 2.0 15 30

Value creation plan

Exit timing

Illustrative. Values at stake and target modes are set with the decision owners; the four decisions come from the open questions of the sponsor and two deal partners.

Reading the example. Add-on go / no-go has the largest value-weighted gap: synergy and price are judged by hand and nothing is learned from past add-ons. The value creation plan follows; it already uses augmented diagnosis and will reuse the synergy analysis. Exit timing has the largest raw gap but less value at stake, so it is staged. Learning is at 0 or J for every decision: decision records come first.

The result of the assessment#

Output What IT says Example (illustrative)
Recommended entry level L1 to L5, from the firm profile (chapter 07) and the decision with the largest value-weighted gap L3: add-on go / no-go as one decision policy, with price as its optimization; L1 decision records for the committee alongside
Audit scope (phase 2) The decisions and components the Audit examines in depth Add-on go / no-go and price; perceive context, counterfactual prediction and learning
Readiness work Enabling actions that must start now so the entry is not blocked later Decision records for the last 14 add-ons; data room access before NDA; clean-team rules
Re-assessment The same questionnaire and heat map, repeated at Scaling Expected: D1 and D5 from 2.4 to 3.5; Learning from 0 to A on the first policy

Workbook · Template

Decision heat map and assessment result#

h

PHASE1 Discovery, again at 6 ScalingCOMPLETED BYvencortex team with decision ownersGATEInput to G1FEEDSTemplate L, Template D, Template R

One row per candidate decision named in the questionnaire. Mark each cell 0, J, S, A or M for how it is done today; set the target mode per column; enter the value at stake. Then compute the gap and gap × value and fill the result.

PERCEIVE CAUSAL COUNTERFACTUAL CONTEXTUAL VALUE AT GAP × CANDIDATE DECISION GOVERNANCE CONTEXT DIAGNOSIS PREDICTION OPTIMIZATION LEARNING STAKE GAP VALUE CANDIDATE DECISION GOVERNANCE PERCEIVE CONTEXT CAUSAL DIAGNOSIS COUNTERFACTUAL PREDICTION CONTEXTUAL OPTIMIZATION LEARNING VALUE AT STAKE GAP GAP × VALUE
Target mode Target mode

0 not done · J judgement only, unstructured · S judgement, structured and recorded · A augmented · M automated. Gap: steps from today to target, summed over the six columns.

Hand-off check for a candidate solution#

WHAT IS PASSED TODAY
LOGGED?
VALUE LOST AT THE HAND-OFF

From stage → to

STAGE Result Recommended entry level L1 to L5 · first decision or analysis · why STAGE Result Recommended entry level L1 to L5 · first decision or analysis · why
Audit scope Decisions and components for phase 2 Readiness work Enabling actions · owner · date (Template R) Audit scope Decisions and components for phase 2 Readiness work Enabling actions · owner · date (Template R)

Part i i i

24. Use case definition at three levels#

A use case is defined as a decision, or as an analysis or a solution around decisions, never as a tool or a dataset. An archetype directory gives the starting point; three templates, one per level, each come with an explanation, a blank form and a worked example.

09

Analysis, decision policy, solution

10

Archetype directory

11

Template A · Analysis use case

12

Template B · Decision use case

13

Defining a solution · Templates T and C

09

Three levels

25. Analysis, decision policy, solution#

The three templates come from the shared framework. A solution holds stages and the set of decisions at each; a decision policy holds one decision and the analyses it combines; an analysis holds four components. They are filled from the bottom up when engineering and from the top down when planning.

C · SOLUTION USE CASE · PE buy and build › › › › 5 Exit 1 Thesis 2 Platform 3 Add-on pipeline 4 Value creation Stage 3 holds a set of decisions: add-on selection · add-on go / no-go and price · add-on financing B · DECISION USE CASE · Add-on go / no-go and price Type Go / No-Go, price and financing C · SOLUTION USE CASE · PE buy and build 1 Thesis › 2 Platform › 3 Add-on pipeline › 4 Value creation › 5 Exit Stage 3 holds a set of decisions: add-on selection · add-on go / no-go and price · add-on financing B · DECISION USE CASE · Add-on go / no-go and price Type Go / No-Go, price and financing by policy optimization · IC decides Goal: platform equity IRR +2 pp per add-on · leverage at most 5.5x Its Recommend step combines three analyses: Synergy analysis Price analysis Financing analysis A · ANALYSIS USE CASE · Synergy capture How much synergy will add-on X deliver, and how fast? Perceive context (automated) · causal diagnosis, counterfactual prediction and contextual optimization (augmented) · output: $1.6m to $6.0m, 5% to 95% REUSE · SCALE AN ANALYSIS The same synergy analysis also serves the value creation plan (stage 4), the exit case (stage 5) and every later add-on of the fund. Each reuse costs a configuration, not a build.
Figure 4 One worked example runs through Part III: a PE buy-and-build platform (Template C), its add-on go / no-go and price decision at stage 3 (Template B) and the synergy analysis it combines (Template A), which is reused by later stages and every later add-on. Illustrative.
Section A · analysis B · decision policy C · solution
Identity Question, archetype, decisions served, owner Question, solution and stage, type, objective, owner, frequency Name and domain, archetype, solution owner
Constraints Regulatory constraints on data, models and conduct Regulatory and mandate constraints Mandate: goals, constraints, regulatory frame
Today Main KPIs, main inefficiencies today, data and source Status quo from the Audit (Template D) Where hand-offs lose value today
Core Four components, each with analysis archetypes, output and human / AI mode Recommendation: options, actions, optimization goal, decision criteria, insights; the analyses it combines Stakeholders; stages and the set of decisions per stage; hand-offs and decision lifecycles
Assets Output handed to the policy; dependencies Causal decision graph, process, decision record Decision assets and aggregated views
Value and proof Proof and reuse; value; readiness Alpha and value, qualification, proof plan Underwritten alpha, proof, roadmap, scale path
Path Scale an analysis Scale a decision policy Scale a solution; roadmap through the maturity levels

Which template first? Planning a programme: C first, as stages with the set of decisions at each, then B for the decisions on the roadmap. Engineering a decision: B first, then A for each analysis it needs. Starting at L2: A first, with a short B for the decision it feeds, because an analysis without a decision cannot show value. In every case, start from an archetype (chapter 10), not from a blank page.

10

Archetype directory

26. Archetype directory#

No use case starts from a blank page. A solution starts from a solution archetype, a decision policy from a policy archetype (its decision policy type and decision archetype), an analysis from the analysis families of each component. The archetype gives the first draft of the template; the client's stages, goals and data make it theirs.

TEMPLATE C

Pick a solution archetype

Take its typical stages and key decisions as the draft for the stage register. Rename and reorder the stages to the client's decision lifecycle: stages are use-case specific.

TEMPLATE B

Pick a policy archetype: type and decision archetype

It sets one of the four decision policy types, the output of the Recommend step and the typical analysis chain, which becomes the draft list of analyses.

TEMPLATE A

Pick analysis families per component

For each of the four components, pick the families and analysis archetypes, and start from their default human / AI mode.

Solution archetypes · for template c solution archetype Typical stages (use-case specific) Key decisions Typical stakeholders
Long / short equity Strategy · Idea selection · Entry and sizing · Monitoring · Exit Long and short selection, conviction test, sizing, hedging, trim and exit triggers PM, analysts, risk
Value and long-only Universe · Idea selection · Portfolio construction · Monitoring · Exit Ranking, margin-of-safety test, position sizing, sell discipline PM, analysts, risk
Multi-asset credit Allocation · Issuer selection · Sizing and hedging · Monitoring · Rebalancing Sleeve allocation, issuer selection, duration and spread positioning, rating migration triggers CIO, sleeve PMs, credit analysts, risk
Private credit Origination · Underwriting · Structuring · Monitoring · Workout Underwriting go / no-go and pricing, covenant design, early warnings, workout route Credit committee, deal team, portfolio monitoring
PE buy and build Thesis · Platform · Add-on pipeline · Value creation · Exit Platform bid, add-on go / no-go and price, financing, value creation levers, exit timing Deal and operating partners, IC, platform CFO
Total portfolio (CIO) Strategic allocation · Tactical allocation · Manager selection · Risk overlay · Review Allocation across asset classes, manager selection, hedging overlay, rebalancing triggers CIO, investment committee, risk
Family office Mandate · Manager and direct deal selection · Allocation · Monitoring · Liquidity Manager true alpha, direct deal go / no-go, allocation, liquidity planning Family principals, CIO, advisers
Corporate capital allocation Strategy · Allocation · M&A · Performance diagnosis · Capital return Capex and R&D allocation, M&A target, treatment of underperformers, dividend and buyback CEO, CFO, board
Solution archetype Typical stages (use-case specific) Key decisions Typical stakeholders
M&A advisory Mandate · Universe · Ranking · Pricing · Process and timing Buyer or target ranking, price, timing of the process Advisory partner, deal team, client board
Activist and active owner Screening · Engagement agenda · Campaign · Monitoring · Exit Target selection, which changes create most value, escalation, exit PM, engagement team, governance experts

Stages are typical starting points, never the framework's own list. Each solution names its stages with the client.

Policy archetypes: decision policy types and decision archetypes · For template b

Type Decision archetype and question Output Typical analysis chain Example policies
Selection Selection Which options are worth our time? Ranked shortlist Investable universe → contextual valuation → contextual returns → ranking and selection → archetype profiles Long idea generation; deal sourcing; issuer selection; manager selection
Go / No-Go Conviction test Do we have the conviction to proceed under these terms? Yes or no, with terms Peers → causal mispricing → default or catalyst prediction → stress testing → Go / No-Go → archetype profiles Add-on acquisition; loan underwriting; fundamental short; M&A target
Policy optimization Price What price should we pay or accept? Price range and walk-away Contextual base rates → contextual valuation → contextual returns → price optimization → contextual impact Bid and walk-away price; loan spread; exit price
Treatment Which operational levers must management pull? Levers in order Peers → causal driver decomposition → fundamental forecasting → treatment optimization → contextual response Value creation plan; engagement agenda; performance turnaround
Timing What is the best window to act? Window to act Similar situation: point in time → contextual regime → catalyst prediction → timing optimization → systemic impact Exit timing; catalyst trade; refinancing window
Financing How should we structure the deal and the capital? Capital structure Contextual regime → macro forecasting → stress testing → financing optimization → systemic response Add-on financing; covenant design; dividend versus buyback
Allocation How should resources be deployed across opportunities? Allocation weights Contextual regime → knowledge fusion → contextual returns → ripple-through → allocation optimization → systemic synergy Strategic and tactical allocation; capex budget; sleeve allocation
Portfolio construction How should we size positions? Position sizes Codependency → causal driver decomposition → contextual returns → portfolio construction → systemic synergy Position sizing; book construction; issuer weights
Hedging Which exposures do we keep, and which do we hedge? Hedge ratios Causal driver decomposition → macro forecasting → stress testing → hedging optimization → systemic response Factor and rate hedge; FX overlay
Monitoring Trigger design What should trigger action? Triggers and thresholds Codependency → anomaly → default or catalyst prediction → ripple-through → monitoring → contextual impact Early warning on borrowers; position review triggers; rating migration watch
Root cause Why did the outcome differ from what we expected? Variance attribution Relative performance → causal driver decomposition → catalyst effect and impact → contextual and systemic impact Post-investment review; manager true alpha; loss review

Analysis families by component · For template a

Component Analysis family Analysis archetypes Default mode Earlier archetype name
Perceive context Context representation Ontology, embedding, causal modelling Augmented
Context scan Similar entity (thematic, causal drivers, behaviour, codependency); similar situation (trajectory, point in time) Automated Peer and similar-asset scan
Context construction Investable universe, sector, theme, peers Augmented Peer and similar-asset scan
Context generation Imputation, synthetic data, contextual base rates Automated Peer and similar-asset scan
Causal diagnosis Valuation and mispricing Contextual valuation, causal mispricing Augmented
Drivers and regime Causal driver decomposition, contextual regime (global, local, macro, micro) Augmented
Standing Relative performance, causal rating, anomaly Automated; causal rating augmented
Counterfactual prediction Prediction and fusion (will) Knowledge fusion, contextual returns, default prediction, fundamental and macro forecasting, catalyst prediction Augmented Catalyst prediction
Simulation (could, would) Stress testing, scenario simulation, ripple-through across levels, catalyst effect and impact Automated; stress design augmented, scenario selection judgement only Catalyst prediction, Scenario and downside
Contextual optimization Selection Ranking and selection Augmented; criteria weights judgement only
Policy optimization Price, treatment, timing, financing, allocation, portfolio construction, hedging Augmented Sizing and risk budget
Monitoring Trigger design, early warnings, root cause Automated; thresholds judgement only Exit and review
Go / No-Go Conviction against threshold and terms Judgement only at the final call
Explanation Attribution, response, archetype (turnaround target, archetype profiles) Automated

11

Template a

27. Template A · Analysis use case#

An analysis use case defines one analysis well enough that it can be built once, reused in many decisions and proven on its own. It names the question, the constraints and today's inefficiencies, the data, and its four components: Perceive context, Causal diagnosis, Counterfactual prediction and Contextual optimization. For each component it records the analysis archetypes, the output and the human / AI mode.

Field guide#

Field identity and purpose What to capture Detail in
A1 · Identity and owner Name, ID, version, the analysis family it starts from (chapter 10); who owns it, keeps it calibrated and signs its output Ch. 10, analysis families
A2 · Analysis question The one question the analysis answers, with its horizon
A3 · Decisions served Every decision policy that uses it (nesting rule), and what each does with the output Template B, analyses
A4 · Output The calibrated result handed to the policy: form, unit, horizon, refresh Template B, analyses
CONSTRAINTS AND TODAY
Regulatory constraints Data, model and conduct rules that apply Template R
Main KPIs Outcome metrics the analysis must improve and its own quality metrics (detailed in A10, A11)
A5 · Main inefficiencies today Status quo and the human limits it removes: bandwidth, memory, noise, bias, slow updating Template D
A6 · Data and source Client proprietary data, DeOS ecosystem data, expert input; point in time, gaps (Template X) Template X
COMPONENTS
A6 · Perceive context Analysis archetypes, output, human / AI mode Template X; context scan
A6 · Causal diagnosis Analysis archetypes, output, human / AI mode Template G
A6 · Counterfactual prediction Analysis archetypes, output, human / AI mode Template E (expert input)
A6 · Contextual optimization Analysis archetypes, output, human / AI mode Template O, K
A7 · Human and AI parts in detail Per analysis archetype where it differs from the component mode, and why; expert questions used (Template E) Template E; manual, section 3
EXPLANATION, PROOF AND
VALUE
A8 · Dependencies Upstream analyses it is conditioned on; downstream analyses and policies that use it
Field What to capture Detail in
A9 · Explanations Attribution, response and archetype explanations offered; what the decision owner must be able to ask Template I
A10 · Proof Base rate, benchmark (human-only and status quo), metric, threshold to pass G2, point-in-time backtest window Templates F, W
A11 · Value Alpha moved, benefit type and benefit equation, with the client's numbers Template V
A12 · Reuse Other decisions, teams or funds that could use it Template S
Readiness and risk Data access, model risk, expert availability, red lines Template R

Test of a good analysis use case. It answers one question, names at least one decision it serves, sets a human / AI mode for each of its four components, states its output as a distribution or probability, and says before it is built how it will be shown to be better than the status quo.

Companion reference. The DeOS Analysis Manual explains how to define, build, chain and prove an analysis. Its analysis template (Part II) uses the same codes A1 to A12 as this template; its section 3 sets out the human and AI parts per analysis archetype, and its Part IV gives each analysis archetype with its output, proof metric and default mode.

Template a · Human / AI

28. Human and AI, component by component#

Every component of every analysis gets one of three modes. The mode says who does the work; it never moves the decision away from its owner.

THE HUMAN / AI CONTINUUM · ONE SCALE, SET PER COMPONENT OF EVERY ANALYSIS Human share of the work AI share of the work Judgement only Augmented Automated People decide; DeOS records the judgment DeOS proposes; experts adjust and are fused DeOS runs it; people watch the exceptions WHO DOES THE WORK THE HUMAN / AI CONTINUUM · ONE SCALE, SET PER COMPONENT OF EVERY ANALYSIS Human share of the work AI share of the work Judgement only People decide; DeOS records the judgment Augmented DeOS proposes; experts adjust and are fused Automated DeOS runs it; people watch the exceptions WHO DOES THE WORK People decide the component. DeOS captures the judgment in a structured form (ranges, confidence, rationale) and records it. WHEN IT FITS No data or precedent, a new situation, or accountability that must stay with a person (mandate trade-offs, final approval). HUMAN ROLE Decides and records the rationale EXAMPLES The final go / no-go of the investment committee; a relationship or integrity assessment WHO DOES THE WORK DeOS proposes and quantifies; experts add, adjust and are fused with the model by their track record. WHEN IT FITS Data and judgment both carry information; the stakes justify a human look. HUMAN ROLE Gives structured predictions, adjusts and signs off EXAMPLES Synergy capture likelihood fused from model and operating partners; contextual valuation WHO DOES THE WORK DeOS runs the component end to end. People set it up, watch its quality and handle exceptions. WHEN IT FITS Data is rich, the task repeats, errors are visible and cheap to correct. HUMAN ROLE Reviews exceptions and quality reports EXAMPLES Connecting a data room; a comparable- firms scan; imputation of missing financials Set per component: Perceive context · Causal diagnosis · Counterfactual prediction · Contextual optimization. Fixed before the analysis runs, calibrated from outcomes. Automate what is high in volume and low in ambiguity; augment where expert knowledge is scarce in the data; keep judgement only where accountability requires a person. The decision itself, at policy level, always stays with the named decision owner.
Figure 5 The human / AI continuum. Judgement only, augmented and automated are one scale, set per component of every analysis and calibrated from outcomes; the decision itself stays with the decision owner.
Mode What IT means When IT fits Human role Example
Automated DeOS runs the component end to end. People set it up, watch its quality and handle exceptions. Data is rich, the task repeats, errors are visible and cheap to correct. Reviews exceptions and quality reports Connecting a data room; a comparable-firms scan; imputation of missing financials
Augmented DeOS proposes and quantifies; experts add, adjust and are fused with the model by their track record. Data and judgment both carry information; the stakes justify a human look. Gives structured predictions, adjusts and signs off Synergy capture likelihood fused from model and operating partners; contextual valuation
Judgement only People decide the component. DeOS captures the judgment in a structured form (ranges, confidence, rationale) and records it. No data or precedent, a new situation, or accountability that must stay with a person (mandate trade-offs, final approval). Decides and records the rationale The final go / no-go of the investment committee; a relationship or integrity assessment

The rule. Automate what is high in volume and low in ambiguity; augment where expert knowledge is scarce in the data; keep judgement only where accountability, values or regulation require a person. The mode is set per component of every analysis (in detail per analysis archetype), before the analysis runs, and calibrated from outcomes. The decision itself, at policy level, always stays with the named decision owner.

Setting the mode#

Start from the default mode of the analysis family (chapter 10)

Is there enough data and precedent? If not, judgement only, structured

Is an error visible and cheap to correct? Then automated can fit

Do data and judgment both carry information? Then augmented

Must accountability stay with a person? Then judgement only

Record today's mode from the heat map (Template H) next to the target

Record the mode per component (A7); where one analysis archetype differs, record it in detail

Never set a mode for the decision itself: it stays with the decision owner

Example · synergy analysis of an add-on#

Component Mode Why
Perceive context Automated Peer scan and base rates repeat for every add-on; errors are visible in the peer set and cheap to correct.
Causal diagnosis Augmented The driver decomposition is proposed by DeOS; operating partners add what the data cannot see.
Counterfactual prediction Augmented Model and four operating partners are fused by their track record on past integrations.
Contextual optimization Augmented; final lever choice judgement only Levers are ranked with their explanation; the operating partner chooses and records why.

Workbook · Template

Analysis use case · blank form#

a

PHASE3 EngineeringCOMPLETED BYAnalysis owner, decision architectGATEG1 draft · G2 proofFEEDSTemplate B, B14

One form per analysis. Reused analyses keep their ID and add the new decision to A3. Tick one human / AI mode per component; record exceptions per analysis archetype in A7. Codes A1 to A12 match the DeOS Analysis Manual.

Identity and purpose
A1 · Identity and owner Name, ID, version, the analysis family it starts from (chapter 10); who owns it, keeps it calibrated and signs its output
A2 · Analysis question The one question the analysis answers, with its horizon
A3 · Decisions served Every decision policy that uses it (nesting rule), and what each does with the output
A4 · Output The calibrated result handed to the policy: form, unit, horizon, refresh
CONSTRAINTS AND TODAY
Regulatory constraints Data, model and conduct rules that apply
Main KPIs Outcome metrics the analysis must improve and its own quality metrics (detailed in A10, A11)
A5 · Main inefficiencies today Status quo and the human limits it removes: bandwidth, memory, noise, bias, slow updating
A6 · Data and source Client proprietary data, DeOS ecosystem data, expert input; point in time, gaps (Template X)
COMPONENTS COMPONENT ANALYSIS ARCHETYPES A7 · HUMAN / OUTPUT (CATALOGUE) AI MODE
A6 · Perceive context Scope · What should we look at, and what is it comparable to? Context engineering: representation, scan, construction, generation Note: Automated Augmented Judgement only
A6 · Causal diagnosis Diagnose · What really drives outcomes here, and where do we stand? Causal diagnosis: valuation, mispricing, driver decomposition, regime, relative performance, rating, anomaly Note: Automated Augmented Judgement only
A6 · Counterfactual prediction Predict · What will, could and would happen? Probabilistic prediction and fusion (will); counterfactual simulation (could, would) Note: Automated Augmented Judgement only
ANALYSIS ARCHETYPES A7 · HUMAN / COMPONENT OUTPUT (CATALOGUE) AI MODE A6 · Contextual optimization Automated Contextual optimization (Selection, policy Recommend · What should we do, given our Augmented optimization, Go / No-Go); explanation goals and constraints, and why? Judgement only Note: A7 · COMPONENT ANALYSIS ARCHETYPES (CATALOGUE) OUTPUT A7 · HUMAN / AI MODE A6 · Contextual optimization Recommend · What should we do, given our goals and constraints, and why? Contextual optimization (Selection, policy optimization, Go / No-Go); explanation Automated Augmented Judgement only Note: A7 · Human and AI parts in detail Per analysis archetype where it differs from the component mode, and why; expert questions used (Template E) EXPLANATION, PROOF AND VALUE A8 · Dependencies Upstream analyses it is conditioned on; downstream analyses and policies that use it A9 · Explanations Attribution, response and archetype explanations offered; what the decision owner must be able to ask A10 · Proof Base rate, benchmark (human-only and status quo), metric, threshold to pass G2, point-in-time backtest window A11 · Value Alpha moved, benefit type and benefit equation, with the client's numbers A12 · Reuse Other decisions, teams or funds that could use it Readiness and risk Data access, model risk, expert availability, red lines
Decision Policy 3 - Demo Analysis Overview Insights Data Room Scan Expert Prediction Pipeline last executed at Oct 12, 2026, 7:15:13 PM byMichael Choe Run Data Room Private Credit Tran… Scan Buyer Scan Scan Stock Selection Scan Data Room Add-On Transaction Expert Prediction Investment Case Due… Expert Prediction Corporate Bond Due … Data Room Portfolio Company Data Room Buyout Transaction Ai Function Capital Allocation … Ai Function Expert Assumptions … Ai Function Assessment Agent
Figure 6 Analysis pipeline. One analysis as a pipeline: data rooms and scans (perceive context) and expert predictions feed AI functions (here capital allocation, expert assumptions, an assessment agent) that produce the result the policy recommends on. A filled Template A becomes this pipeline: its perceive context (data rooms and scans), its expert input and its AI functions, set by the human / AI mode of each component.

Workbook · Template a · Worked example

Synergy capture of add-on X A I L LU STRATIVE WORKBOOK · TEMPLATE A · WORKED EXAMPLE A Synergy capture of add-on X I L LU STRATIVE

Example: PE buy and build. Names, numbers and thresholds are illustrative, not client data or product output.

Identity and purpose
A1 · Identity and owner Synergy capture · A-01 · v1.0 · archetype: catalyst prediction (synergy capture) with catalyst effect and impact · owner: operating partner; associate runs it; model validation by risk
A2 · Analysis question How much synergy will add-on X deliver, and how fast?
A3 · Decisions served Add-on go / no-go; value creation plan; exit case
A4 · Output Synergy distribution: expected $4.1m run-rate EBITDA, 5% to 95%: $1.6m to $6.0m, by month 18
CONSTRAINTS AND TODAY
Regulatory constraints Clean-team rules for competitor data before closing; GDPR for customer lists; model validation
Main KPIs Calibration of synergy forecasts; synergies captured against case; days from NDA to synergy case
A5 · Main inefficiencies today Synergies set top-down by an optimistic deal team; no base rates; past integrations not reused
A6 · Data and source Target data room and CRM overlap (client); private company financials and PE transactions (DeOS); operating partners (experts)
Components component Analysis archetypes (catalogue) Output A7 · human / AI mode
A6 · Perceive context Scope · What should we look at, and what is it comparable to? Similar entity: codependency; peers; contextual base rates Peer set of 30 comparable add-ons; base rate: 58% of planned synergies captured by month 18 Automated Augmented Judgement only
A6 · Causal diagnosis Diagnose · What really drives outcomes here, and where do we stand? Causal driver decomposition of the target's margins; relative performance Target's EBITDA margin 3 pp below contextual peers; two thirds from procurement and pricing Automated Augmented Judgement only
A6 · Counterfactual prediction Predict · What will, could and would happen? Catalyst prediction (synergy capture) fused with operating partners; catalyst effect and impact on platform EBITDA Synergy distribution fused with four operating partners: expected $4.1m, 5% to 95%: $1.6m to $6.0m Automated Augmented Judgement only
A6 · Contextual optimization Recommend · What should we do, given our goals and constraints, and why? Treatment optimization of integration levers; contextual synergy explanation Ranked integration levers (procurement, pricing, back office) with timing and explanation Automated Augmented Judgement only Final lever choice judgement only
A7 · Human and AI parts in detailPeers and base rates automated; synergy capture prediction augmented, four operating partners answer two expert questions; final choice of integration levers judgement only, recorded with its reason

Explanation, proof and value

  • A8 · Dependencies · Upstream: peer scan and target financials · downstream: add-on price analysis, value creation plan
  • A9 · Explanations · Synergy attributed to procurement, pricing and back office; response of synergy to integration speed; three past add-ons as reference cases
  • A10 · Proof · Backtest on the last 14 add-ons, point in time: forecast error against the deal team's synergy cases; pass G2 if error is 30% lower and intervals cover 80% to 90% of outcomes
  • A11 · Value · Analysis and behavioural alpha: overpayment for synergies avoided = add-ons per year × synergy overestimate × multiple paid
  • A12 · Reuse · Every add-on policy of the fund; the value creation plan and exit case; the second platform of fund IV
  • Readiness and risk · Data room after NDA; clean team in place before signing; panel of four operating partners; no red lines

12

Template b

29. Template B · Decision use case#

A decision use case defines one decision as a decision policy: its type, objective, owner and constraints, the recommendation it makes (options, actions, optimization goal, decision criteria, insights), the analyses its Recommend step combines, and its process and record. Filled in completely, it is the specification of a policy in DeOS.

Field guide#

Field identity What to capture Detail in
B1 · Decision question The one question this policy answers
B2 · Solution and stage Where the decision sits in the decision lifecycle Template C
B3 · Type One of four decision policy types (Selection, Policy optimization, Go / No- Go, Monitoring) and its decision archetypes Ch. 01, four types; ch. 10
B4 · Objective What a good decision achieves, in measurable terms
B5 · Owner Who recommends and who decides; who signs off
B6 · Frequency and trigger How often the decision is made and what starts it
CONSTRAINTS
B7 · Regulatory constraints Rules the decision must satisfy, and how they are checked Template R
B8 · Mandate constraints Limits inherited from the solution mandate Template C, mandate
RECOMMENDATION
B9 · Options The alternatives analysed in full depth, including waiting and passing Template O
B10 · Actions What the recommendation tells people to do, with size and timing Template O
B11 · Optimization goal The measurable outcome the options are ranked on, with expected value and interval Template O
B12 · Decision criteria Standardized measures of conviction, grouped into perspectives Template K
B13 · Insights Observations in the standard grammar that explain the criteria Template I
ANALYSES
B14 · Analyses The analyses the policy combines, each with its own template; the policy's Recommend step combines them Template A, one per analysis
B15 · Combination How the Recommend step combines them: Selection, Policy optimization, Go / No-Go Ch. 01, four types
B16 · Causal decision graph Levers, externals, intermediates, outcomes and goals the analyses feed Template G
Field process and record What to capture Detail in
B17 · Process Stages, steps, roles and sign-offs inside the policy Template P
B18 · Hand-offs What the policy receives and hands on Template C
B19 · Decision record and proof What is logged, and the KPIs the policy is judged on Templates N, U
TODAY, VALUE AND
QUALIFICATION
B20 · Status quo How the decision is made today, by whom, and what it costs Template D
B21 · Alpha and value Which of the four alphas move; benefit equation with the client's numbers Template V
B22 · Qualification and entry D · V · F · R at policy level, potential and complexity, entry mode Templates L, R
B23 · Proof plan Analysis, outcome and process evidence for the Prove gate Templates F, W

Test of a good decision use case. It has a named owner, an objective that can be measured, a known frequency, at least two real options, an optimization goal written before the analyses are chosen, and a stated way the Recommend step combines them. A use case framed as "use AI for research" fails all six.

Workbook · Template

Decision use case · blank form#

b

PHASE2 Audit (identity, status quo), 3 Engineering (the rest)COMPLETED BYDecision owner, decision architectFEEDSTemplate C, stage register

Identity, constraints and status quo are filled in the Audit; recommendation, analyses and process in Engineering.

IDENTITY B1 · Decision question The one question this policy answers B2 · Solution and stage Where the decision sits in the decision lifecycle B3 · Type One of four decision policy types (Selection, Policy optimization, Go / No-Go, Monitoring) and its decision archetypes B4 · Objective What a good d IDENTITY B1 · Decision question The one question this policy answers B2 · Solution and stage Where the decision sits in the decision lifecycle B3 · Type One of four decision policy types (Selection, Policy optimization, Go / No-Go, Monitoring) and its decision archetypes B4 · Objective What a good decision achieves, in measurable terms B5 · Owner Who recommends and who decides; who signs off B6 · Frequency and trigger How often the decision is made and what starts it CONSTRAINTS B7 · Regulatory constraints Rules the decision must satisfy, and how they are checked B8 · Mandate constraints Limits inherited from the solution mandate RECOMMENDATION B9 · Options The alternatives analysed in full depth, including waiting and passing B10 · Actions What the recommendation tells people to do, with size and timing B11 · Optimization goal The measurable outcome the options are ranked on, with expected value and interval B12 · Decision criteria Standardized measures of conviction, grouped into perspectives B13 · Insights Observations in the standard grammar that explain the criteria
ANALYSES B14 · Analyses The analyses the policy combines, each with its own template; the policy's Recommend step combines them B15 · Combination How the Recommend step combines them: Selection, Policy optimization, Go / No-Go B16 · Causal decision graph Levers, externals, intermediates, outcomes an ANALYSES B14 · Analyses The analyses the policy combines, each with its own template; the policy's Recommend step combines them B15 · Combination How the Recommend step combines them: Selection, Policy optimization, Go / No-Go B16 · Causal decision graph Levers, externals, intermediates, outcomes and goals the analyses feed ANALYSES REGISTER ANALYSIS A ID NEW OR REUSED FEEDS (CRITERION, OPTION OR ACTION) MODES (SC · CD · CP · CO) PROCESS AND RECORD B17 · Process Stages, steps, roles and sign-offs inside the policy B18 · Hand-offs What the policy receives and hands on B19 · Decision record and proof What is logged, and the KPIs the policy is judged on TODAY, VALUE AND QUALIFICATION B20 · Status quo How the decision is made today, by whom, and what it costs B21 · Alpha and value Which of the four alphas move; benefit equation with the client's numbers B22 · Qualification and entry D · V · F · R at policy level, potential and complexity, entry mode B23 · Proof plan Analysis, outcome and process evidence for the Prove gate

Workbook · Template b · Worked example

Add-on go / no-go and price B I L LU STRATIVE WORKBOOK · TEMPLATE B · WORKED EXAMPLE B Add-on go / no-go and price I L LU STRATIVE

Example: PE buy and build. Names, numbers and thresholds are illustrative, not client data or product output.

Identity
B1 · Decision question Should we acquire add-on X, and on what terms?
B2 · Solution and stage PE buy and build · stage 3 Add-on pipeline
B3 · Type Go / No-Go, with price and financing set by policy optimization · decision archetypes: conviction test, price, financing
B4 · Objective Raise platform equity IRR by at least 2 pp per add-on within the leverage limit
B5 · Owner Deal partner recommends; investment committee decides; operating partner signs the integration plan
B6 · Frequency and trigger 8 to 12 add-ons a year; started by a signed NDA
CONSTRAINTS
B7 · Regulatory constraints Merger control thresholds; clean-team rules before signing; LPA concentration limit; AI regulation classification
B8 · Mandate constraints Leverage at most 5.5x; add-on at most 25% of platform EBITDA
RECOMMENDATION
B9 · Options Acquire at ask · acquire at walk-away price · wait two quarters · pass
B10 · Actions Bid range, financing mix, first three integration levers, review triggers
B11 · Optimization goal Platform equity IRR: expected, 5% and 95% bounds
B12 · Decision criteria Performance, strategic fit, potential; e.g. synergy capture at least 60% likely
B13 · Insights Target's cash conversion cycle 12 days longer than its contextual peers
ANALYSES
B14 · Analyses Synergy, price and financing analysis
B15 · Combination Go / No-Go on conviction, then price and financing optimization
B16 · Causal decision graph Price, financing mix, integration levers → revenue and cost synergies, leverage → platform EBITDA → equity IRR
Analyses register analysis A ID New or reused Feeds (criterion, option or action) Modes (sc · CD · CP · co)
Synergy analysis A-01 New; reused by value creation Criterion synergy capture likelihood; integration actions M·A·A·A
Price analysis A-02 New Bid range and walk-away price; option ranking M·A·A·A
Financing analysis A-03 Reused from platform bid Financing mix; leverage constraint M·M·A·A
Process and record
B17 · Process Perceive the context, run the analyses, recommend, IC decision
B18 · Hand-offs Receives the shortlist from add-on selection; hands the synergy case to value creation
B19 · Decision record and proof Recommendation, conviction, override and reason; add-on IRR against case
TODAY, VALUE AND QUALIFICATION
B20 · Status quo Deal team builds the case in spreadsheets; synergies set top down; the committee sees one point estimate
B21 · Alpha and value Analysis and behavioural alpha: 0.5 turn lower average multiple on $4m add-on EBITDA × 10 add-ons = $20m less paid
B22 · Qualification and entry D 8 · V 9 · F 6 · R 4 → potential 17, complexity 8 · land now · parallel run on the next three add-ons
B23 · Proof plan Calibration of synergy forecasts; add-on IRR against case; days from NDA to committee

13

Defining a solution

30. Defining a solution: process, decision map and Template C#

A solution is the decision lifecycle of one domain: use-case specific stages, each holding the decisions made at it, run by named stakeholders under one mandate. Defining one means finding where in the decision lifecycle each decision is made, by whom and on what, before any policy is engineered. Six steps do that; the decision map (Template T) records the result and feeds the solution use case (Template C).

Six steps from decision lifecycle to solution#

Step What to do Output
1 Draw the decision lifecycle Pick the closest solution archetype and write the client's own stages in order, with what starts and ends each stage. Four to seven stages; never a generic list. Stage list with entry and exit per stage
2 Find the decision points Walk each stage with its owner and mark every moment where someone commits resources, passes or stops a case, sets a term, allocates, or changes their mind. Note what triggers each (an event, a cadence, a signal) and how often it occurs. Decision points per stage, each with trigger and frequency
3 Name and type each decision At each decision point, name the decision as a decision policy: its objective, its type (Selection, Policy optimization, Go / No-Go or Monitoring), who decides and who is involved. A point with two questions holds two policies. One named decision policy per point, with owner
4 Trace inputs and hand-offs For every decision, write what it needs from earlier decisions and what it passes on; mark the hand-offs that lose value today (re-work, delay, judgement without evidence). Group decisions whose outcomes interact into decision lifecycles optimized together. Hand-offs, decision lifecycles and the value lost between decisions
5 Rate today and target Take each decision's mode today from the decision heat map and set the target mode per component; estimate the value at stake per decision; name the analyses and assets several decisions share. Mode today and target, value at stake, decision assets
6 Order the roadmap Rank the decision policies by value at stake and readiness; decide which land first (the entry level), which are staged and which stay as they are. Carry the result into the solution template, one decision use case per policy on the roadmap. Roadmap; solution template filled; decision use cases opened

Who does it. The solution owner walks the decision lifecycle with one decision owner per stage and the decision architect; steps 1 to 4 take one workshop per stage in Discovery, steps 5 and 6 use the heat map and the audit. The map is redrawn at Scaling, when stages and decisions have changed.

DECISION MAP · PE BUY AND BUILD · WHICH DECISION AT WHICH POINT › › › › 5 Exit 1 Thesis 2 Platform 3 Add-on pipeline 4 Value creation Sector and thesis Platform screening Add-on selection Value creation levers Exit timing and route Selection Selection Selection Policy opt. Policy opt. Trigger: annua DECISION MAP · PE BUY AND BUILD · WHICH DECISION AT WHICH POINT 1 Thesis › Sector and thesis Selection Trigger: annual 2 Platform › Platform screening Selection Trigger: per target Platform bid and price Go / No-Go Trigger: per bid 3 Add-on pipeline › Add-on selection Selection Trigger: monthly Add-on go / no-go, price Go / No-Go Trigger: per add-on Add-on financing Policy opt. Trigger: per add-on 4 Value creation › Value creation levers Policy opt. Trigger: 100-day, yearly Early warning Monitoring Trigger: continuous 5 Exit Exit timing and route Policy opt. Trigger: quarterly Post-investment review Monitoring Trigger: per exit SEQUENCE · platform bid, its financing and the first add-on are optimized as one path · HAND-OFFS (green) are logged human decisions · Roadmap: add-on go / no-go lands first (L3)
Figure 7 A decision map for a PE buy-and-build platform: the stages across the top, the decision points beneath each with the decision made there, its type and its trigger, and the hand-offs between stages. The add-on go / no-go lands first and is the worked example of Templates B and A. Illustrative.

Where a decision point hides#

A stage rarely names its decisions. Walk it with its owner and look for five moments; each is a decision point and gets a row in the map.

LOOK FOR THE MOMENT Commitment Money, people or time are committed or withdrawn: a bid, an allocation, a hire, a budget. Pass or stop A case moves on, is parked or is killed: screening, IC approval, a covenant waiver. LOOK FOR THE MOMENT Commitment Money, people or time are committed or withdrawn: a bid, an allocation, a hire, a budget. Pass or stop A case moves on, is parked or is killed: screening, IC approval, a covenant waiver.

Terms

A price, size, structure, timing or route is set: offer price, leverage, exit window.

Watch

A signal is checked against a trigger and the owner must decide whether to act: early warning, thesis break, rebalancing.

Review

An outcome is judged and the next cycle changes: post-investment review, mandate review.

Test of a good decision map. Every stage has at least one decision point; every point has a trigger, a named decision of one type and one owner; every decision says what it needs and what it hands on; the hand-offs that lose value today are marked; and the roadmap column says which decisions land first. A map in which every decision is "judgement only, as today" is a description, not a definition.

Workbook · Template

Decision map · blank form#

t

PHASE1 Discovery (draft), 2 Audit (complete)COMPLETED BYSolution owner, decision owner per stage, decision architectGATEG1 Discover · G4 Scale (update)FEEDSTemplate C, C10 and stage register; one Template B per decision; Template L

One row per decision point. Stages in order; a stage can have several rows. Decision: name and type (Selection, Policy optimization, Go / No-Go, Monitoring). Mode today from the heat map (Template H); target per component. Roadmap: the entry level and order in which the decision policies land.

DECISION DECIDES NEEDS MODE DECISION · TODAY → STAGE POINT · · FROM HANDS ON ROADMAP TYPE TRIGGER INVOLVED EARLIER TARGET STAGE DECISION POINT · TRIGGER DECISION · TYPE DECIDES · INVOLVED NEEDS FROM EARLIER HANDS ON MODE TODAY → TARGET ROADMAP

1

2

3

4

5

Workbook · Template t · Worked example

WORKBOOK · TEMPLATE T · WORKED EXAMPLE
T
PE buy-and-build platform, fund IV
I L LU STRATIVE

Example: PE buy and build. Names, numbers and thresholds are illustrative, not client data or product output.

Stage Decision point · trigger Decision · type Decides · involved Needs from earlier Hands on Mode today → target Roadmap
1 Thesis Fund strategy set · annual, at fund raise Sector and thesis selection · Selection Managing partner · partners, LP advisory Prior fund outcomes and root causes Thesis criteria and target profile to platform screening Judgement only → Augmented L4 (later)
2 Platform Target enters screening · per target Platform screening · Selection Deal partner · associates Thesis criteria Shortlist with conviction to the bid decision Judgement only → Augmented L4
2 Platform IC meeting on a shortlisted target · per bid Platform bid go / no-go and price · Go / No-Go, price by Policy optimization IC · deal partner, sector expert Shortlist conviction; thesis constraints Platform case, maximum price and integration capacity to the add-on pipeline Judgement only → Augmented L3, second
3 Add-on pipeline Add-on sourced · per target, monthly pipeline review Add-on selection · Selection Deal partner · platform CEO Platform case and integration capacity Ranked add-on shortlist to the add-on go / no-go Judgement only → Automated scan, augmented ranking L3, with the go / no-go
3 Add-on pipeline IC meeting on an add-on · per add-on Add-on go / no-go and price · Go / No-Go, price by Policy optimization IC · deal partner, operating partner Shortlist; synergy case; integration capacity left Synergy case and price to financing and to the value creation plan Judgement only → Augmented L3, first
3 Add-on pipeline Price agreed · per add-on Add-on financing · Policy optimization Deal partner, platform CFO · lenders Price; leverage path to date Debt per add-on and the leverage path to value creation and exit Augmented → Augmented L3, with the go / no-go
4 Value creation 100-day plan and annual budget · per add-on, then yearly Value creation levers · Policy optimization (treatment) Operating partner · platform CEO and CFO Synergy case; price paid Levers, targets and the EBITDA bridge to monitoring and the exit case Judgement only → Augmented L4
4 Value creation Monthly reporting; trigger fires · continuous Early warning on the value creation plan · Monitoring Operating partner · platform CFO Levers and targets Triggers to the operating partner and IC; outcomes to the decision memory Not done → Automated, judgement on action L4
5 Exit Exit window review · quarterly from year 3 Exit timing and route · Policy optimization (timing) Managing partner, IC · deal partner, advisers Realised EBITDA bridge; market regime Exit decision and proceeds to the fund view Judgement only → Augmented L4
5 Exit Deal closed · per exit Post-investment review · Monitoring (root cause) Managing partner · deal and operating partners Every decision record of the platform Findings back to thesis, selection and price Not done → Augmented L5

One row per decision point. The decision column becomes a decision use case (one policy template each); the hands-on column fills the solution template's hand-offs; the roadmap column orders the decision portfolio.

Example: PE buy and build, illustrative.

13

Template c

31. Template C · Solution use case#

A solution use case is the end-to-end template: the decision lifecycle of one domain, organized as use-case specific stages with the set of decisions at each stage, run by named stakeholders under one mandate. It is the level at which hand-offs between decisions and the decision lifecycle are optimized, and the level of maturity L4 and L5. Its stage register is the decision map (Template T) condensed to one row per stage.

Field guide#

Field identity What to capture Detail in
C1 · Name and domain The strategy, fund, asset class or programme the solution covers
C2 · Solution archetype The archetype it starts from (directory, chapter on archetypes) Ch. 10, solution archetypes
C3 · Solution owner Who owns the solution end to end and signs off its roadmap
MANDATE
C4 · Goals What the solution must achieve, measurably
C5 · Constraints Limits every decision must respect: risk, leverage, liquidity, concentration, ESG, regulation
C6 · Regulatory frame Regulation that applies to the whole solution Template R
STAKEHOLDERS
C7 · Sponsor and owners Sponsor, solution owner, decision owner per stage Template B, owner
C8 · Contributors Analysts, experts, data and technology
C9 · Control and recipients Risk, compliance, auditors, investors, lenders
STAGES AND DECISIONS
C10 · Stages The use-case specific stages, in order. Not a generic list Stage register
C11 · Decisions per stage The set of decision policies at each stage, each with its own template Template B, one per decision
C12 · Hand-offs What each decision passes to the next; each hand-off is a logged human decision Template N
C13 · Sequences Decisions optimized together as one path
DECISION ASSETS AND VIEWS
Field What to capture Detail in
C14 · Decision assets The predefined, standardized components every policy reuses so the decision runs the same way every time: policy templates, data ontologies, metrics, criteria libraries, causal decision graphs, expert prediction questions, data connectors, reusable analyses Templates A, G, K, X
C15 · Aggregation Views across all decisions of the solution
VALUE, PROOF AND ROADMAP
C16 · Underwritten alpha Benefit equations per stage, with the client's own numbers Template V
C17 · Proof Analysis, outcome, process and innovation proof Templates F, W
C18 · Roadmap Which policies land first, which are staged; maturity level targeted Template S
C19 · Scale path Next platform, fund or strategy; a total portfolio view; the team building in DeOS Studio
C20 · Risks and readiness Solution-level red lines and enabling actions Template R

Test of a good solution use case. Every stage has its set of decisions and every decision an owner; every hand-off says what is passed and is logged; decision assets are named, so the next policy reuses them; and the value of connecting decisions is underwritten separately from the value of each decision.

Workbook · Template

Solution use case · blank form#

c

PHASE2 Audit (draft), 3 Engineering, 6 Scaling (update)COMPLETED BYSolution owner, sponsor, decision architectFEEDSTemplate S, roadmap

Start in Discovery with the stages and the decisions at each; complete the register as decision use cases are defined.

IDENTITY C1 · Name and domain The strategy, fund, asset class or programme the solution covers C2 · Solution archetype The archetype it starts from (directory, chapter on archetypes) C3 · Solution owner Who owns the solution end to end and signs off its roadmap MANDATE C4 · Goals What the solution m IDENTITY C1 · Name and domain The strategy, fund, asset class or programme the solution covers C2 · Solution archetype The archetype it starts from (directory, chapter on archetypes) C3 · Solution owner Who owns the solution end to end and signs off its roadmap MANDATE C4 · Goals What the solution must achieve, measurably C5 · Constraints Limits every decision must respect: risk, leverage, liquidity, concentration, ESG, regulation C6 · Regulatory frame Regulation that applies to the whole solution STAKEHOLDERS C7 · Sponsor and owners Sponsor, solution owner, decision owner per stage C8 · Contributors Analysts, experts, data and technology C9 · Control and recipients Risk, compliance, auditors, investors, lenders STAGES AND DECISIONS C10 · Stages The use-case specific stages, in order. Not a generic list C11 · Decisions per stage The set of decision policies at each stage, each with its own template C12 · Hand-offs What each decision passes to the next; each hand-off is a logged human decision C13 · Sequences Decisions optimized together as one path

Stage register

STAGE
DECISIONS AT THIS STAGE
(POLICY · TYPE)
DECISION
OWNER
B ID
HANDS ON TO
(POLICY · TYPE) OWNER DECISION ASSETS AND VIEWS C14 · Decision assets The predefined, standardized components every policy reuses so the decision runs the same way every time: policy templates, data ontologies, metrics, criteria libraries, causal decision graphs, expert prediction questions, data co (POLICY · TYPE) OWNER DECISION ASSETS AND VIEWS C14 · Decision assets The predefined, standardized components every policy reuses so the decision runs the same way every time: policy templates, data ontologies, metrics, criteria libraries, causal decision graphs, expert prediction questions, data connectors, reusable analyses C15 · Aggregation Views across all decisions of the solution VALUE, PROOF AND ROADMAP C16 · Underwritten alpha Benefit equations per stage, with the client's own numbers C17 · Proof Analysis, outcome, process and innovation proof C18 · Roadmap Which policies land first, which are staged; maturity level targeted C19 · Scale path Next platform, fund or strategy; a total portfolio view; the team building in DeOS Studio C20 · Risks and readiness Solution-level red lines and enabling actions

Workbook · Template c · Worked example

WORKBOOK · TEMPLATE C · WORKED EXAMPLE
C
PE buy-and-build platform, fund IV
I L LU STRATIVE

Example: PE buy and build. Names, numbers and thresholds are illustrative, not client data or product output.

Identity
C1 · Name and domain Buy-and-build platform in fragmented B2B services, fund IV
C2 · Solution archetype PE buy and build
C3 · Solution owner Managing partner, with the head of value creation
MANDATE
C4 · Goals Fund net IRR above 20%; 2.5x MOIC on the platform
C5 · Constraints Leverage at most 5.5x; no single add-on above 25% of platform EBITDA; fund LPA exclusions
C6 · Regulatory frame AIFMD reporting; merger control; AI regulation classification of the decisions
STAKEHOLDERS
C7 · Sponsor and owners Managing partner; deal partners (stages 1 to 3); operating partner (stage 4); IC (all approvals)
C8 · Contributors Associates, sector experts, platform CFO, data team
C9 · Control and recipients Compliance officer, LP advisory committee, lenders' agent
STAGES AND DECISIONS
C10 · Stages 1 Thesis · 2 Platform · 3 Add-on pipeline · 4 Value creation · 5 Exit
C11 · Decisions per stage Stage 3: add-on selection, add-on go / no-go and price, add-on financing
C12 · Hand-offs Add-on go / no-go hands the synergy case and price to the value creation plan
C13 · Sequences Platform bid, financing and first add-on as one sequence
Stage register stage Decisions at this stage (policy · type) Decision owner B ID Hands on to
1 Thesis Sector and thesis selection · Selection Managing partner B-01 Thesis criteria to platform screening
2 Platform Platform bid go / no-go and price · Go / No-Go; policy optimization (price) Deal partner; IC B-02 Platform case and integration capacity to the add-on pipeline
3 Add-on pipeline Add-on selection · Selection; add-on go / no-go and price · Go / No-Go, policy optimization (price); add-on financing · Policy optimization Deal partner; IC B-03 to B-05 Synergy case and price to the value creation plan
4 Value creation Value creation levers · Policy optimization (treatment); early warnings · Monitoring (trigger design) Operating partner, platform CFO B-06, B- 07 Realised EBITDA bridge to the exit case
5 Exit Exit timing and price · Policy optimization (timing, price) Managing partner; IC B-08 Outcomes to root cause and the next thesis
Decision assets and views
C14 · Decision assets Services ontology; data room connectors; synergy and price analyses reused for every add-on
C15 · Aggregation Platform value bridge by add-on and lever; fund view by vintage
VALUE, PROOF AND ROADMAP
C16 · Underwritten alpha 0.5 turn lower average add-on multiple across 10 add-ons
C17 · Proof Calibration of synergy forecasts; add-on outcomes against case
C18 · Roadmap Add-on go / no-go first (L3), then the full decision lifecycle (L4)
  • C19 · Scale path · Second platform of fund IV on the same assets; then a fund view across platforms (L5)
  • C20 · Risks and readiness · Integration capacity of the platform team; data room access before signing; no red lines

Part IV

32. Engineering templates#

The sheets that turn a decision use case into an engineered decision policy. Each produces a reusable decision asset: an audit, a causal decision graph, goals and actions, a criteria library, insight and expert question specifications, a data room and a process.

14

D · Decision audit

15

G · Causal decision graph canvas

16

O · Goals, options and actions

17

K · Perspectives and criteria

18

I · E · Insights and expert questions

19

X · Context and data room

20

P · N · Process and decision record

14

Template d

33. Decision audit#

What does the decision cost today, and what could it earn? The audit describes the status quo, diagnoses inefficiencies by the four alphas and quantifies the augmentation and automation potential.

Decision audit D PHASE 2 Audit COMPLETED BY Decision owner with the vencortex team GATE G1 Discover FEEDS Template B, B20; Template V D Decision audit PHASE 2 Audit COMPLETED BY Decision owner with the vencortex team GATE G1 Discover FEEDS Template B, B20; Template V
Status quo
Decision and owner Template B ID · who decides today
Current framework Steps, inputs, templates, committees, criteria actually used
Time and cost Hours per decision by role · cycle time · decisions per year
Track record What is known about outcomes of past decisions of this type
INEFFICIENCY DIAGNOSIS ALPHA TYPICAL SYMPTOMS OBSERVED IN THIS DECISION
Governance Inconsistent process, missing records, overrides not reviewed
Information Data collected by hand, slow, incomplete; peers by habit
Analysis Single-point forecasts, few scenarios, no causal model
Behavioural Anchoring, overconfidence, disposition effect, groupthink

Potential

STEP OR ANALYSIS
AUGMENT (HUMAN +
AI)
AUTOMATE
BEYOND HUMAN LIMITS:
WHAT BECOMES POSSIBLE
AUGMENT (HUMAN + BEYOND HUMAN LIMITS: AUTOMATE AI) WHAT BECOMES POSSIBLE AUGMENT (HUMAN + AI) AUTOMATE BEYOND HUMAN LIMITS: WHAT BECOMES POSSIBLE

Step or analysis

15

Template g

34. Causal decision graph canvas#

Start at the goals on the right and work backwards. Keep the boundary tight: a node belongs in the graph only if it would change the decision.

Causal decision graph canvas#

g

PHASE3 EngineeringCOMPLETED BYDecision owner, analysts, decision architectGATEG2FEEDSTemplate B, B16; Template K
EXTERNALS · outside our control LEVERS INTERMEDIATES INTERMEDIATES OUTCOMES GOALS what we control how the effect travels what we are measured on targets on outcomes AI · Expert · Rule · Qual. AI · Expert · Rule · Qual. AI · Expert · Rule · Qual. AI · Expert · Rule · Qual. AI · Expert · Rule · Qual. EXTERNALS · outside our control LEVERS what we control AI · Expert · Rule · Qual. AI · Expert · Rule · Qual. AI · Expert · Rule · Qual. AI · Expert · Rule · Qual. INTERMEDIATES how the effect travels AI · Expert · Rule · Qual. AI · Expert · Rule · Qual. AI · Expert · Rule · Qual. AI · Expert · Rule · Qual. INTERMEDIATES AI · Expert · Rule · Qual. AI · Expert · Rule · Qual. AI · Expert · Rule · Qual. AI · Expert · Rule · Qual. OUTCOMES what we are measured on AI · Expert · Rule · Qual. AI · Expert · Rule · Qual. AI · Expert · Rule · Qual. GOALS targets on outcomes AI · Expert · Rule · Qual. AI · Expert · Rule · Qual. AI · Expert · Rule · Qual. Fill right to left (why) to design; read left to right (how) to decide. Mark each node's evidence source.
Figure 8 Blank canvas. Write one node per box; mark its evidence source: AI prediction, expert estimate, business rule or qualitative judgment. Add arrows and feedback loops by hand.

Node register

NODE
TYPE
EVIDENCE
FED BY (ANALYSIS ·
TEMPLATE A ID)
CRITERION OR GOAL IT
INFORMS

Good-practice check

  • Goals agreed before any node was added · Every node names its evidence source
  • Every node would change the decision · Feedback loops are drawn explicitly
  • Levers and externals are separated · Graph reused across cases; changed only through root-cause findings

16

Template o

35. Goals, options and actions#

Goals turn the desired outcome into something measurable; options are the alternatives analysed in full depth; actions are the concrete settings to execute.

Goals, options and actions#

o

PHASE3 EngineeringCOMPLETED BYDecision ownerGATEG2FEEDSTemplate B, B9–B11

Optimization goals

GOAL
TYPE
METRIC · CDG
OUTCOME
TARGET
HORIZON AND PREFERENCE
(TRADE-OFF)
OUTCOME (TRADE-OFF) Primary: Secondary: Secondary: Constraint: Constraint: Type: primary goal (optimized), secondary goal (traded off with explicit preference), constraint (hard limit, never breached). OPTIONS # OPTION WHAT IT IS GOOD FOR WHAT IT GIVES UP 1 2 3 4 Always include waiting and passing w OUTCOME (TRADE-OFF) Primary: Secondary: Secondary: Constraint: Constraint: Type: primary goal (optimized), secondary goal (traded off with explicit preference), constraint (hard limit, never breached). OPTIONS # OPTION WHAT IT IS GOOD FOR WHAT IT GIVES UP 1 2 3 4 Always include waiting and passing where they are real options. The best-ranked option can still carry low conviction. ACTIONS ACTION (LEVER) RANGE UNIT WHAT-IF QUESTIONS THE DECISION MAKER WILL ASK
WHAT-IF QUESTIONS THE DECISION ACTION (LEVER) RANGE UNIT MAKER WILL ASK ACTION (LEVER) RANGE UNIT WHAT-IF QUESTIONS THE DECISION MAKER WILL ASK

17

Template k

36. Perspectives and criteria library#

Criteria express conviction from 0 to 100%, always in the same direction, so they can be compared, weighted and fused into perspectives and the decision conviction score.

Perspectives and criteria library#

k

PHASE3 EngineeringCOMPLETED BYDecision owner, analysts, riskGATEG2FEEDSTemplate B, B12

Perspectives

PERSPECTIVE QUESTION IT ANSWERS FOR THIS DECISION WEIGHT OR PREFERENCE Attractiveness Risk Fit Timing Execution · Expert view PERSPECTIVE QUESTION IT ANSWERS FOR THIS DECISION WEIGHT OR PREFERENCE Attractiveness Risk Fit Timing Execution · Expert view

Criteria

CRITERION
PERSPECTIVE
TYPE
MEASURE ·
CONVERSION
THRESHOLD
WEIGHT
CDG NODE

Measure

Types: cross-sectional performance · performance over time · catalyst likelihood · catalyst impact · expert judgment · exposure · fixed business rule · dynamic business rule. Time: historic, current or predicted. A threshold turns a criterion into a gate. Scale: 0% impossible · 5% exceptionally unlikely · 15% very unlikely · 30% unlikely (red) · 40% somewhat unlikely · 59% about even chance (amber) · 70% somewhat likely · 85% likely · 95% very likely · 100% almost certain (green).

18

Templates i and e

37. Insights and expert questions#

Insights explain criteria in a standard grammar a decision maker can check. Expert questions turn judgment into a calibrated, fusable signal.

Insight specification I PHASE 3 Engineering COMPLETED BY Analysts GATE G2 FEEDS Template B, B13; Template K I Insight specification PHASE 3 Engineering COMPLETED BY Analysts GATE G2 FEEDS Template B, B13; Template K
METRIC
WHEN
WHAT
HOW
MEASURED
HOW MOVING
WHY AND WHAT NEXT

When: historic · current · future. What: over time · point-in-time · distribution · composition · relationship. How measured: absolute · relative to history · relative cross-sectional. How moving: trend · velocity · momentum · seasonality · volatility · anomaly. Why: contextual or systemic impact, synergy or response.

Expert question card E PHASE 4 Implementation / PoV COMPLETED BY Decision architect with experts GATE G2 FEEDS Template A, A7 Question One question, precisely worded, with horizon Input type Potential · risk · horizon · target assumption · return hurdle · time to target · effect size, duration, like E Expert question card PHASE 4 Implementation / PoV COMPLETED BY Decision architect with experts GATE G2 FEEDS Template A, A7 Question One question, precisely worded, with horizon Input type Potential · risk · horizon · target assumption · return hurdle · time to target · effect size, duration, likelihood · cost · synergy timing · feedback · preference Answer format Most likely value with 5% and 95% bounds, or one of five scales: probability 0–100% · positive judgment 0–10 · negative judgment 0–10 · confidence 5–100% · uncertainty 1– 100% Colour Red, amber or green per answer: green supports the decision, red speaks against it; on negative judgments a high value is red Panel Who answers (internal, external) · minimum number of answers
Base rate shown Contextual base rate from the context scan shown at the moment of judgment Resolution How and when the answer is checked against the outcome, for calibration feedback Base rate shown Contextual base rate from the context scan shown at the moment of judgment Resolution How and when the answer is checked against the outcome, for calibration feedback

19

Template x

38. Context and data room#

As much context as necessary, as little as possible. The data room starts from the decision, not from the data, and the foundation grows with every decision.

Context and data room X PHASE 3 Engineering COMPLETED BY Data and technology, analysts GATE G1 readiness · G2 FEEDS Template A, A6 data and components X Context and data room PHASE 3 Engineering COMPLETED BY Data and technology, analysts GATE G1 readiness · G2 FEEDS Template A, A6 data and components

Concepts, metrics and sources

Point

CONCEPT
METRICS
SOURCES
INT. /
EXT.
POINT
IN
TIME
STATUS

Int. /

ONTOLOGY AND CONTEXT SCAN Objects and granularity Company, business unit, product line, asset, deal, person · geography, sector, theme · frequencies Context scan Similar entities and / or similar situations · similarity by characteristics, causal drivers, behavior, theme · trajectory read Context ge ONTOLOGY AND CONTEXT SCAN Objects and granularity Company, business unit, product line, asset, deal, person · geography, sector, theme · frequencies Context scan Similar entities and / or similar situations · similarity by characteristics, causal drivers, behavior, theme · trajectory read Context generation Where data is scarce: simulated comparables or counterfactual situations, labelled as generated Gaps and enabling actions Missing data, access, integration, owner and date

20

Templates p and n

39. Process, governance and the decision record#

The process makes sure every decision is made the same way and nothing is missing. The decision record keeps what was known, what was recommended and what was decided, point in time.

Process and governance#

p

PHASE3 EngineeringCOMPLETED BYDecision owner, risk and complianceGATEG2FEEDSTemplate B, B17
TASK OR STEP
ROLE
TRIGGER
DEADLINE
REQUIRED INPUT ·
GUARDRAIL
Sign-offs Who signs off, above which limits Guardrails Mandate limits, investment policy and rules encoded; what blocks submission Override rule How an override is recorded and when it is reviewed Sign-offs Who signs off, above which limits Guardrails Mandate limits, investment policy and rules encoded; what blocks submission Override rule How an override is recorded and when it is reviewed
Decision record N PHASE 4 Implementation / PoV, 5 Run COMPLETED BY Decision owner GATE G2 · G3 FEEDS Proof plan; Template V realised N Decision record PHASE 4 Implementation / PoV, 5 Run COMPLETED BY Decision owner GATE G2 · G3 FEEDS Proof plan; Template V realised
Decision Policy and version · asset · date · decision level Recommendation Conviction · perspectives · ranked options · actions · goal probabilities Probability and threshold at decision time Probability of meeting the goals · policy threshold · expected loss if the thesis fails · what would change Decision Policy and version · asset · date · decision level Recommendation Conviction · perspectives · ranked options · actions · goal probabilities Probability and threshold at decision time Probability of meeting the goals · policy threshold · expected loss if the thesis fails · what would change the call Final decision Option chosen · actions set · override? reason
Expectations to track Signals and dates that would confirm or break the thesis Outcome, judged against the distribution Filled later: outcome vs. goals · which scenario occurred and its weight at decision time · lever, external or broken link · verdict (Template U) Expectations to track Signals and dates that would confirm or break the thesis Outcome, judged against the distribution Filled later: outcome vs. goals · which scenario occurred and its weight at decision time · lever, external or broken link · verdict (Template U)

Part v

40. Delivery templates: qualify, underwrite, gate, prove#

The sheets that govern the portfolio of use cases: which ones to land and which to stage, what each is expected to deliver, whether it is ready, how it passes each gate and how it is proven.

21

L · Long list and D · V · F · R scoring by level

22

V · Alpha underwriting

23

R · Readiness check

24

Q · Gate review

25

F · S · Proof plan and roadmap

26

W · U · Evaluation by level and decision evaluation

21

Template l

41. Long list and D · V · F · R scoring by level#

Every candidate is scored 0 to 10 on desirability, viability and feasibility; risk is scored separately because it can veto a use case that looks attractive on everything else. Each candidate is scored at its own level: a solution, a decision policy or an analysis.

Long list and scoring L PHASE 1 Discovery COMPLETED BY Sponsor, decision owners, vencortex GATE G1 FEEDS Template S L Long list and scoring PHASE 1 Discovery COMPLETED BY Sponsor, decision owners, vencortex GATE G1 FEEDS Template S
USE CASE LEVEL D V F R POT. COMPL. DECISION · OWNER USE CASE LEVEL D V F R POT. COMPL. DECISION · OWNER

Level: analysis (Template A), decision policy (Template B) or solution (Template C). Potential = D + V. Complexity = (10 − F) + R. Decision: land now, stage, bundle, park. A red-line risk stops a use case regardless of its scores.

SCORE HIGH (8–10) WHEN

Strategic fit, named sponsor, clear advantage over status quo, fits how the team works, value visible quickly

Several alphas at stake, high frequency, large capital per decision, fast and durable value

Data available, few integrations, decision easy to structure, team capability (D3, D4, D6 high)

Regulatory exposure, dependency, low AI output quality or trust, weak user control (high = more risk)

SCORE LOW (0–3) WHEN

No sponsor, the team does not feel the problem, value only visible after a long time

Rare decision, small stakes, value hard to measure

Data missing, heavy integration, decision poorly defined

Well-understood, controllable, low exposure

The questions differ by level#

The same four dimensions are asked differently for a solution, a decision policy and an analysis. Use the column of the candidate's level.

Solution Decision policy Analysis
Is the decision lifecycle a strategic priority with a named sponsor? Do the stakeholders want one connected way of deciding? Does the decision owner want it? Is the decision frequent and painful enough to matter? Do analysts and experts want the relief? Does it answer a question they ask every time?
Value across all decisions, including what is lost at hand-offs today. Decisions per year × capital per decision × Δ decision quality, plus efficiency and risk. Better accuracy and calibration, time saved, and how many policies reuse it.
A shared ontology and data across stages; integration with the systems each stage uses. Can options, goals and criteria be made explicit? Is the data for its analyses in reach? Is the data there, of usable quality, and does an archetype exist to start from?
Concentration of change in one programme; regulatory scope of the whole strategy. Regulatory classification, model risk, user control and trust in the recommendation. Output quality, explainability, data protection and conduct rules on the inputs.

D · Desirability

V · Viability

F · Feasibility

R · Risk

Roll-up rules#

Score each candidate at its own level: a solution, a decision policy or an analysis.

An analysis reused by several policies counts its value in each of them; reuse raises viability.

A red-line risk at any level stops everything above it: a blocked analysis blocks the policy that depends on it.

The level you land decides the maturity entry: an analysis lands at L2, a policy at L3, a solution at L4 or L5.

20 LAND NOW · low-hanging fruit STAGE · moonshots, level by level POTENTIAL = Desirability + Viability 10 BUNDLE · quick wins, ride along PARK · waste of resources 0 0 10 20 COMPLEXITY = (10 − Feasibility) + Risk LAND NOW · low-hanging fruit STAGE · moonshots, level by level BUNDLE · quick wins, ride along PARK · waste of resources 0 0 10 10 20 20 COMPLEXITY = (10 − Feasibility) + Risk POTENTIAL = Desirability + Viability
Figure 9 Blank quadrant. Plot each use case by its computed potential and complexity, never by hand placement; label it with its long-list row and level (A, B or C).

22

Template v

42. Alpha underwriting#

Underwrite before engineering; compare at the Realisation gate. Write every benefit as an equation with the client's own inputs, never as a fixed promise.

Alpha underwriting V PHASE 1–2 underwrite, 5 Run realised COMPLETED BY Sponsor, decision owner, vencortex FEEDS Template B, B21; Template C, C16 V Alpha underwriting PHASE 1–2 underwrite, 5 Run realised COMPLETED BY Sponsor, decision owner, vencortex GATE G1 · G3 FEEDS Template B, B21; Template C, C16
Benefit Governance Information gate g1 · g3 Analysis Behavioural
Efficiency
Speed
Scale
Coverage
Effectiveness
Risk reduction
Decision quality
Experience
Transparency
Learning
Innovation
Fundraising success

The twelve rows are the measures under the six forms of benefit of the delivery framework: efficiency (with speed), scale, coverage, effectiveness (with risk reduction, decision quality and fundraising success), experience (with transparency, learning and fundraising success) and innovation.

Benefit equations

UNDERWRITTEN REALISED (G3) UNDERWRITTEN REALISED (G3)
BENEFIT EQUATION AND INPUTS Effectiveness Efficiency Coverage Risk Decision quality BENEFIT EQUATION AND INPUTS Effectiveness Efficiency Coverage Risk Decision quality

Effectiveness = decisions per year × capital per decision × Δ decision quality (bps). Efficiency = hours freed × decisions per year × cost per hour. Risk = Δ probability of a loss event × exposure × average loss. Coverage = additional opportunities × hit rate × value per opportunity. Decision quality = calibration, policy adherence and conviction against outcome, measured per decision type.

Benefits arrive at different speeds. Efficiency, speed and transparency show within weeks; decision quality within months at the analysis level; effectiveness and fundraising success over years. Template W plans the evidence by level so the fast benefits are shown first while the slow ones build their record.

23

Template r

43. Readiness check#

A use case can be attractive and still not ready. Readiness is checked at every gate; when it is missing, defer and enable, never force.

Readiness check R PHASE Every gate COMPLETED BY Sponsor, data and technology, compliance GATE G1–G4 FEEDS Template Q R Readiness check PHASE Every gate COMPLETED BY Sponsor, data and technology, compliance GATE G1–G4 FEEDS Template Q
AREA Check Status Enabling action · owner · date
People Named decision owner and executive sponsor; champions; time for training; willingness to record conviction and rationale
Process Decision mapped (steps, roles, inputs, approvals); outcome measurable; frequency known; KPIs and underwritten alpha agreed; feedback loop
Technology Access to the data the decision needs; integration points; dedicated instance and security set-up; data ontology
Compliance Regulatory classification incl. AI regulation; model risk and validation; data protection; mandate and guardrails encoded

Status: READY moves on · ENABLE moves on with a named action tracked to the next gate · BLOCKED stays on the roadmap; another use case takes its slot.

24

Template q

44. Gate review#

A gate is a decision about a decision. Each has a question, required evidence and a fixed set of outcomes. No use case moves on because of momentum.

Gate review Q PHASE End of each stage COMPLETED BY Gate board GATE G1–G4 FEEDS Template S Use case and gate ID · level (analysis, decision policy, solution) · gate · date · board members Q Gate review PHASE End of each stage COMPLETED BY Gate board GATE G1–G4 FEEDS Template S Use case and gate ID · level (analysis, decision policy, solution) · gate · date · board members

G1 · Discover

Is this decision worth engineering?

D, V, F, R scores, potential and complexity
Decision audit with underwritten alpha
Named owner and sponsor
Readiness check

Outcome: Go to Prove · Stage · Park

G2 · Prove

Does it work on our own decisions?

Analysis proof: calibration and accuracy

Outcome proof: point-in-time backtest, parallel run

Process proof: adherence, cycle time, audit

Trust: output quality, trust calibration, user control

Evidence planned by level (Template W); probabilities recorded at decision time (Template N)

Human / AI mode per component confirmed against calibration and overrides

Trust: output quality, trust calibration, user control

Outcome: Go to production · Iterate · Stop

G3 · Realisation

Is the underwritten alpha delivered in daily use?

Realised vs. underwritten on the benefit grid

Adoption and usage by decision owners

Overrides and their outcomes

Decision quality: calibration held; losses on above-threshold decisions reviewed as expected cost, not as failures (Template U)

Overrides and their outcomes

Subjective factors rated by decision owners (Template

Learning: improving with outcomes?

W)

Outcome: Keep and grow · Re-engineer · Retire

G4 · Scale

What do we scale, and how far?

Reusable assets: analyses, policies, ontologies
Demand from other teams, funds, strategies
Readiness of the next teams
Team capability to build in DeOS Studio

Outcome: Scale an analysis · decision policy · solution

Decision and conditions

Outcome chosen · conditions · enabling

actions · next review

25

Templates f and s

45. Proof plan, roadmap and decision portfolio#

Proof comes in four kinds; the roadmap runs the use cases as a portfolio, landing the fruit first so it funds and de-risks the moonshots. The seven steps that derive the roadmap with the client, from long list to signed decision portfolio, are in the Solutions Catalogue, chapter 01; Template S records the result.

Proof plan F PHASE 4 Implementation / PoV COMPLETED BY Decision owner, vencortex GATE G2 · G3 · G4 FEEDS Template Q F Proof plan PHASE 4 Implementation / PoV COMPLETED BY Decision owner, vencortex GATE G2 · G3 · G4 FEEDS Template Q
Kind of proof Question Evidence planned · metric · threshold Data and period
Analysis Is each analysis right, and better than the status quo?
Outcome Would and do decisions turn out better, at decision level and over the decision lifecycle?
Process Is the decision made better?
Innovation What can the firm do now that it could not before?
Trust Output quality, trust calibration, user control
Subjective Conviction, uncertainty awareness, explainability, auditability, adjustability, rated by decision owners

Outcome evidence is read against the probabilities recorded at decision time, never against the single outcome alone; see chapter 26.

Roadmap and decision portfolio#

s

PHASE1 Discovery, 6 ScalingCOMPLETED BYSponsor, vencortexFEEDSTemplate C, C18
WAVE
MATURITY
LEVEL
USE CASES (L ROW · A, B,
C ID)
ENTRY MODE
ASSETS REUSED ·
CREATED
LEVEL C ID) 1 2 3 4 5 LEVEL C ID) 1 2 3 4 5

26

Templates w and u

46. Evaluation by level and decision evaluation#

Decision optimization recommends the action with the best expected value within the mandate when its probability of success clears the policy threshold. A bet with good odds still loses some of the time. Judging that loss as a failure is outcome bias, and it leads a firm to abandon sound policies and to promote lucky ones.

One worked example#

A credit policy approves when the probability that the loan meets its return hurdle is at least 60 per cent and sizes exposure to conviction. For one loan the system, fusing the deal team's assessment with its causal model, records 72 per cent and an expected loss of 8 per cent of par if the thesis fails. The committee approves at the recommended size. Eighteen months later the borrower defaults and the position loses 9 per cent.

The loss proves that the failure scenario was real. It cannot show that 72 per cent was wrong: that scenario always carried 28 per cent of the weight, and one draw cannot distinguish a 72 per cent decision from a 40 per cent one. What can is calibration: across all decisions the policy labelled 70 to 75 per cent, did about seven in ten succeed? If so, the loss was the expected cost of a sound policy and the firm should decide the same way again. If only four in ten did, the probabilities were inflated and the policy needs correcting regardless of how this loan turned out. The reverse case matters as much: a loan approved at 45 per cent against the 60 per cent threshold that happens to pay off is a bad decision with a good outcome, and a decision record that holds the probability at the time makes that visible.

HOW TO READ Bad break Earned success Sound: probability at decision time SOUND DECISION Above threshold, calibrated, lost. Above threshold, calibrated, won. cleared the policy threshold and Expected cost of a sound policy. The policy worked. the probability bucket is calibrated. The outcome decides Bad break Above threshold, calibrated, lost. Expected cost of a sound policy. Earned success Above threshold, calibrated, won. The policy worked. Just deserts Below threshold or inflated, lost. Policy breach or miscalibration. Dumb luck Below threshold or inflated, won. Do not promote it. BAD O UTC O M E GOOD OUTCOME SOUND DECISION UNSOUND DECISION HOW TO READ Sound: probability at decision time cleared the policy threshold and the probability bucket is calibrated. The outcome decides the column; the record decides the row. Only the row is a verdict on the decision and the policy.
Figure 10 The verdict on a decided case. The outcome decides the column; the decision record and the calibration of its probability bucket decide the row. Only the row is evidence about the decision and its policy.

Three levels, three speeds of evidence#

Fund performance is a sequence of hundreds of connected decisions, each built from analyses. An analysis repeats hundreds of times a year and its calibration can be checked within weeks; a decision type repeats dozens of times and shows within months; a fund outcome arrives once per cycle and takes years to separate from luck. Template W plans the evidence at each level, on both the qualitative improvement and the outcome, and adds the subjective factors decision owners experience, which decide whether the system is used at all.

Workbook · Template

Evaluation plan by level W PHASE 4 Implementation / PoV, updated at 5 Run COMPLETED BY Decision owner, vencortex GATE G2 · G3 FEEDS Template Q, Template F WORKBOOK · TEMPLATE W Evaluation plan by level PHASE 4 Implementation / PoV, updated at 5 Run COMPLETED BY Decision owner, vencortex GATE G2 · G3 FEEDS Template Q, Template F

One sheet per decision or solution use case. Fill the three levels before the parallel run starts; fill the subjective factors with the decision owners before and during the pilot.

Evidence by level

LEVEL
QUALITATIVE IMPROVEMENT
(WHAT CHANGES IN HOW IT
IS DONE)
OUTCOME EVIDENCE ·
METRIC · THRESHOLD
REPEATS
PER YEAR
FIRST
EVIDENCE
BY
Is done) by
Analysis
Decision
Decision lifecycle
Analysis: drivers and sources covered, causal explanation, uncertainty stated; calibration, Brier score, hit rate, interval coverage, point in time. Decision: policy adherence, conviction recorded, alternatives considered, overrides explained; value against the counterfactual decision in a parallel run. Decision lifecycle: no loss at hand-offs, consistent policy across the decision lifecycle, decision memory reused; attribution of results to decisions, drivers and luck over several cycles. SUBJECTIVE FACTORS, RATED BY THE DECISION OWNERS (1 TO 5) FACTOR WHAT TO ASK BEFORE PILOT OWNER COMMENT
Conviction
Uncertainty awareness
Explainability and auditability
Customizability and adjustability
Trust in the output

Conviction: does it sharpen conviction, and does size follow it? Uncertainty awareness: distributions, not points; chance separated from ignorance; novelty flagged. Explainability: every recommendation traced to drivers, data, assumptions and policy. Customizability: your thesis, criteria and constraints, set and changed by you. Record how often recommendations are overridden and compare the outcomes of followed and overridden recommendations.

Workbook · Template

Decision evaluation U PHASE 5 Run, at every outcome review COMPLETED BY Decision owner with the vencortex team GATE G3 FEEDS Template N; Template V realised; policy update WORKBOOK · TEMPLATE U Decision evaluation PHASE 5 Run, at every outcome review COMPLETED BY Decision owner with the vencortex team GATE G3 FEEDS Template N; Template V realised; policy update

One sheet per decided case whose outcome is reviewed. Read the decision record first; the outcome is the last field, not the first.

Decision record Template N ID · policy and version · asset · date · decision level Probability and threshold at the time Probability of meeting the goals · policy threshold · conviction · expected loss if the thesis fails · size taken Policy followed? Above threshold · guardrails met · override and Decision record Template N ID · policy and version · asset · date · decision level Probability and threshold at the time Probability of meeting the goals · policy threshold · conviction · expected loss if the thesis fails · size taken Policy followed? Above threshold · guardrails met · override and its reason Outcome Realized result vs. goals · which scenario occurred · its weight at decision time Calibration of the bucket Decisions in the same probability bucket to date · share that succeeded · calibrated, inflated or deflated Verdict Earned success · bad break · just deserts · dumb luck Consequence None (sound policy, expected cost) · adjust probabilities or weights · change threshold or criteria · root cause along the causal decision graph

Rule: a loss on an above-threshold decision in a calibrated bucket is the expected cost of the policy and changes nothing but the record. A win on a below-threshold decision is a policy breach and is reviewed as one. Calibration, not the single outcome, is what updates the policy.

Part VI

47. From workbook to DeOS#

A filled template is not a document that sits in a folder. Each one becomes a screen in DeOS, where the solution, its decision policies and their analyses run, are recorded and learn. This part shows where each template lands.

27

Where each template lands in the app

27

In the app

48. Where each template lands in the app#

Five screens carry most of what the templates capture. A consultant who fills Templates A, B and C, with the engineering sheets behind them, has written the first version of each screen. Screens show demo data; names such as Decision Policy 4 are placeholders of the demo instance.

Screen Filled from What appears
Solution Template C: C10 stages, C11 decisions per stage, C7–C9 stakeholders One column per stage, one card per decision policy, with its open decisions, age and progress
Policy overview Template B: B14 analyses, B11 optimization goal; Template A: A6 data and source; Template X Context, the analyses on the prediction side, and the recommendation they feed
Process Template P and B: B17 process; Template X data contexts; Template E expert predictions The steps of the policy, stage by stage, each with an owner
Insights Templates I and E; Template A: A7 human and AI parts (expert input) Expert questions with ranges and confidence, beside the insights and benchmarks for the same metric
Recommendation Template B: B9–B13 recommendation; Templates O and K Ranked options with conviction, the goal with its interval, and the recommended actions

Solution#

FILLED FROM
Template C, solution use case (chapter 13), with its stage register.

C10 Stages become the columns, in the client's own order and names. C11 The set of decisions per stage becomes the policy cards; each card is one Template B. C7 to C9 stakeholders own the cards and sign the hand-offs (C12).

fields

Demo Solution STAGE 1Strategy STAGE 2Idea Selection STAGE 3Portfolio Construction STAGE 4New Stage Add decision Add decision Add decision Add decision 0 Decision Policy 1 Decision Policy 124 d Stage 29/11 0 Decision Policy 9 Decision Policy 924 d Stage 12/11 10 Decision Policy 8 Decision Policy 86 d Stage 13/11 30 Decision Policy 4 Decision Policy 421 d Stage 29/11 0 Decision Policy 3 Decision Policy 319 d Stage 210/11
Figure 11 Solution. A solution in DeOS: use-case specific stages (here Strategy, Idea Selection, Portfolio Construction), each holding the decision policies made at that stage. Each card shows the policy's open decisions, age and progress.

Decision policy overview#

FILLED FROM
Template B, decision use case (chapter 12), and the Template A of each analysis it combines.

B14 The analyses appear on the prediction side, each one a Template A. A6 Data and source and Template X fill the context column. B11 The optimization goal and its interval sit in the recommendation panel; B15 says how the Recommend step combines the analyses.

fields

Decision Policy 4 Overview Process Context Prediction Insights Recommendation ContextPredictionRecommendation Company Context Decision Policy 1 - Portfolio Company Portfolio Company Decision Policy 4 - Private Credit Transaction Private Credit Transaction Decision Policy 4 - Buyout Transaction Buyout Transaction Decision Policy 4 - Add-On Transaction Add-On Transaction Decision Policy 4 - Buyer Scan Buyer Scan Decision Policy 4 - Stock Selection Scan Stock Selection Scan Decision Policy 4 - Corporate Bond Due Diligence Corporate Bond Due Diligence Decision Policy 4 - Investment Case Due Diligence Investment Case Due Diligence Decision Policy 4 - Demo Analysis Demo Analysis Open Open Open OPTION Portfolio OPTIMIZATION GOAL Net Operating Profit After Tax Yield (NOPAT Yield) - Target % 36.58 5% CI 70.16 % EXPECTED % 94.16 95% CI
Figure 12 Decision policy. A decision policy at a glance: its context (data contexts), its analyses, and the recommendation that combines them, with the optimization goal and its 5% to 95% interval.

Process#

FILLED FROM
Template P, process and governance (chapter 20), and the record of Template N.

B17 Each task or step of Template P becomes a step with its role and owner: define the strategic objectives (B4), connect the data contexts (Template X), scan for comparables (perceive context of Template A), incorporate expert predictions (Template E). Sign-offs and guardrails gate the steps; the decision record (Template N) is kept at the end.

fields

Decision Policy 4 Template is in use Unpublish Duplicate Recommendation Insights Prediction Context Process Stage 1Context Stage 2Prediction Define Strategic Objectives Connect Data Context Private Credit Transaction Connect Data Context Buyout Transaction Connect Data Context Portfolio Company Connect Data Context Add-On Transaction Scan Data Context Buyout Transaction Buyer Scan Scan Data Context Add-On Transaction Stock Selection Scan Incorporate Expert Predictions Private Credit Transaction Corporate Bond Due Diligence
Figure 13 Process. The process of a decision policy, stage by stage: define the strategic objectives, connect the data contexts, scan for comparables, incorporate expert predictions. Each step has an owner.

Insights and expert assessment#

FILLED FROM
Template I, insight specification, and Template E, expert question card (chapter 18).

Each expert question card becomes one question with its answer format (minimum, expected, maximum and confidence) and its panel. The insight specification fills the panel beside it: historic values and benchmarks for the same metric, so the base rate is shown at the moment of judgment. The human / AI mode of the counterfactual prediction (A7) decides whether experts are fused with the model or decide alone.

fields

Decision Policy 4 - Investment Case Due Diligence Launch Questions Insights Experts Assessment Cash Conversion Cycle - Long Term Target Assumptions Decision Policy 4 - Buyout Transaction Estimate the target related to Cash Conversion Cycle of Decision Policy 4 - Buyout Transaction Minimum(days) 42 0100 Expected(days) 49 0100 Maximum(days) 59 0100 How confident are you? 85 very confident 85 % Assess Complete 4 / 8 QUESTIONS Assistant Historic Values Benchmarks Working Capital - Historic Stability Time series line VISUALISATIONSNAPSHOT * CONTEXTS Wed Oct 07 2026 Decision Policy 4 - Buyout Transaction Decision Polic Indicators Category 1 Category 2 $ 100M$ 80M$ 60M$ 40M$ 20M$ 0$ -20M
Figure 14 Insights and expert assessment. Insights next to judgment: an expert estimates a target range and confidence for one question, with the historic-values insight and benchmarks for the same metric beside it.

Recommendation#

FILLED FROM
Template B recommendation fields (chapter 12), with Templates O and K behind them.

B9 Options become the ranked list with their conviction. B10 Actions appear grouped by driver family. B11 The optimization goal shows its expected value and 5% / 95% interval. B12 Decision criteria and B13 insights explain the conviction.

fields

Decision Policy 4 Overview Process Context Prediction Insights Recommendation How should we allocate? Wed Oct 07 2026 Portfolio 73 1 Decision Policy 4 - Privat...Private Credit Transaction 91 2 Decision Policy 1 -...Portfolio Company 77 3 Decision Policy 4 - Add-...Add- On Transaction 66 4 Decision Policy 4 - Buyou...Buyout Transaction 47 1 Decision Policy 4 - Private Credit Transaction Private Credit Transaction 90.94% Very Likely - 90.94 % GOALNet Operating Profit After Tax Yield (NOPAT Yield) - Target 36.575% CI 70.24 %EXPECTED 89.5895% CI ACTIONS All Value Drivers Investments Valuation Value Creation Returns Financial Market Dynamics Macroeconomic Dynamics Market Dynamics IncreaseSocial Score - Action ACTION70.77 % IncreaseLong Position - Action ACTION66.2 % IncreaseESG Score - Action ACTION67.51 %
Figure 15 Recommendation. The recommendation of a decision policy: ranked options with their conviction, the selected option's probability of reaching the goal, the goal with its expected value and 5% / 95% interval, and the recommended actions.

Reference

49. Template index#

Code Template Phase Completed by Lands in DeOS Ch.
M Maturity questionnaire (Version I investors, Version E enterprises) and scoring sheet 1 Discovery, 6 Scaling All roles; scored by the vencortex team 06–07
H Decision heat map and assessment result 1 Discovery, 6 Scaling vencortex team with decision owners 08
A Analysis use case 3 Engineering Analysis owner, decision architect Analysis builder, Insights 11
B Decision use case (decision policy) 2 Audit, 3 Engineering Decision owner, decision architect Policy overview, Recommendation 12
T Decision map: stages, decision points, decisions 1 Discovery, 2 Audit Solution owner, decision owner per stage, decision architect Solution 13
C Solution use case (end to end) 2 Audit, 3 Engineering, 6 Scaling Solution owner, sponsor, decision architect Solution 13
D Decision audit 2 Audit Decision owner, vencortex 14
G Causal decision graph canvas 3 Engineering Decision owner, analysts Causal decision graph 15
O Goals, options and actions 3 Engineering Decision owner Recommendation 16
K Perspectives and criteria library 3 Engineering Decision owner, risk Recommendation 17
I · E Insight specification and expert question card 3 Engineering, 4 Implementation / PoV Analysts, experts Insights 18
X Context and data room 3 Engineering Data and technology, analysts Context, data connectors 19
P · N Process and governance; decision record 3 Engineering, 5 Run Decision owner, risk and compliance Process 20
L Long list and D · V · F · R scoring by level 1 Discovery Sponsor, decision owners, vencortex 21
V Alpha underwriting 1 to 2, again at 5 Sponsor, decision owner, vencortex 22
R Readiness check Every gate Sponsor, data and technology, compliance 23
Q Gate review End of each stage Gate board 24
F · S Proof plan; roadmap and decision portfolio 4 Implementation / PoV, 6 Scaling Sponsor, vencortex 25
W · U Evaluation plan by level; decision evaluation 4 Implementation / PoV, 5 Run Decision owner, vencortex 26

Phases: 1 Discovery · 2 Audit · 3 Engineering · 4 Implementation / PoV · 5 Run · 6 Scaling. Gates: G1 Discover · G2 Prove · G3 Realisation · G4 Scale.

© vencortex®. This handbook and workbook describes how DeOS is structured and delivered. It is not investment advice.

← Delivery frameworkGlossary →

Decision optimization with DeOS · vencortex® – DeOS | Decision Optimization System. Converted from DeOS_Compendium_Light.pdf by scripts/pdf_to_markdown.py; built by scripts/build_book_site.py.