Evaluation memory architecture.
A short orientation to the structural elements that distinguish DiligencePilot from a checklist or a prompt library: the criteria anatomy, the nine scoring blocks, the five-section library structure and the discipline-engineering that produced them.
Figure · M.1 · Criterion Anatomy and the 1,000-Criteria Universe — one criterion, located inside fifty modules
Every criterion has a standard structure.
- Criterion number and name — a stable identifier inside the library.
- Applicability tag — ALL, context-specific or profile-specific.
- Deal-breaker status — YES or NO; deal-breaker criteria can independently turn an investment decision negative.
- Test question — the concrete question asked of the file.
- Evidence to look for — what would constitute a credible answer.
- Evidence tier tag — claim / document / external verification (per the Evidence Ladder).
200 criteria each; ten modules per section.
Foundational readings
Founder, product, market, customer, revenue, unit economics, sales channels, production, data and financial health.
Cap table & deep-review
Valuation, campaign demand, anchor, and deep-review layers (product, production, financial model, GTM, feasibility).
Second-pass controls
Equity structure, management, campaign psychology, customer/market/price validation, institutional fit, exit, portfolio fit.
Operational & integrity
Supply, certification, IP, false-traction detection, delivery benchmarks, growth feasibility, profitability bridge, statement depth.
Final gates
Investor psychology and final-gate controls — equity, valuation, demand, PR, management, resilience, competitor response, exit, committee quality.
Decision domains, not a single rubric.
Module scores roll up via context-specific weights into nine decision domains. The final score reads across nine dimensions rather than one.
Market & Competition
Investor relevance, market sizing, competitive positioning.
Product & Technology
Product maturity, technical readiness, IP and defensibility.
Revenue & Unit Economics
Revenue model, pricing, unit economics and traction reality.
Financial Health
Financial discipline, reporting, projection and runway.
Operations & Delivery
GTM, production, sales channels, supply, resilience.
Team & Governance
Founder quality, management, execution discipline, committee quality.
Valuation & Cap Table
Valuation fairness, dilution, equity structure, exit scenarios.
Context-Specific Core
The dominant question of the context (campaign, M&A fit, IPO readiness).
Public Footprint & Risk
External signal quality, PR risk, false-traction detection, expert gate.
Common questions about how the platform works.
A concise reference for reviewers, committee members and integration partners assessing whether DiligencePilot fits their decision flow.
Q01 What does "governance-locked diligence" actually mean?
Once context, profile and the operating constitution are established, the evaluation framework enters Governance-Lock status. Context cannot change. Profile cannot update. The extension list is fixed. The same file is read at the same quality every time, by any reviewer. Reproducibility is the design goal.
Q02 How is DiligencePilot different from a checklist or prompt library?
DiligencePilot is an architecture, not a collection of prompts. Evaluation rules, evidence segmentation, decision logic and delivery discipline are isolated into distinct layers, each governed by an operating constitution. The 1,000 criteria are organised across 50 modules in five sub-libraries (KR-1 to KR-5), with explicit applicability tags, deal-breaker status and evidence-tier requirements.
Q03 What is criterion anatomy?
Every criterion in the library has a standard structure: a stable identifier (e.g. KR-1 · M03 · C-007), an applicability tag (ALL contexts or context/profile-specific), a deal-breaker status (yes/no), a weight block assignment, a test question, evidence to look for, and an evidence tier classification (claim / document / external verification). This anatomy makes every assertion traceable.
Q04 What are the nine scoring blocks?
The nine scoring blocks are: Market & Competition, Product & Technology, Revenue & Unit Economics, Financial Health, Operations & Delivery, Team & Governance, Valuation & Cap Table, Context-Specific Core, and Public Footprint & Risk. The final score is read across all nine — not condensed into a single rubric — because different decision tables care about different blocks.
Q05 What is the Evidence Ladder?
Five evidence tiers segment every claim in a file: Tier 01 Claim (what the founder asserts), Tier 02 Document (shown in the file), Tier 03 External Verification (confirmed against public sources), Tier 04 Contradiction Check (delta between claim, document and external data), Tier 05 Decision-Critical Signal (committee-grade evidence). Confidence rises bottom to top.
Q06 What does "signal-loss prohibition" mean?
A risk or PARTIAL area flagged in early analysis remains visible as an analytical signal even when the delivery language is simplified for the committee. Risk markers persist through every station — analysis, scoring, packaging, executive report, decision label and audit trail. Simplification happens in language, never in signal.
Q07 Is DiligencePilot a substitute for legal or financial advice?
No. DiligencePilot is decision support, not a decision maker. It is not legal advice, not investment advice, not statutory audit, not a binding regulatory opinion. The binding investment decision always rests with the authorised committee. Outputs are produced under explicit scope perimeters and confidentiality discipline.
Q08 What output formats does DiligencePilot produce?
Nine output layers grouped into three clusters: pre-decision (internal criteria score, contradiction/risk-signal map, external query set), decision-grade (final decision note, committee package, committee memo) and post-decision archival (word-ready text, DD request list, institutional Word file). All deliverables are produced under format discipline matching institutional document standards.