Risk analysis for teams that need a source-backed, severity-ranked register they can defend.

A register draws on the offering pack, financials, governing agreements, exposure schedules, and the regulatory record. Analysts rank by instinct; rows lose their evidence; a critical flag arrives with no citation behind it. A risk that cannot be cited is not evidence — it is opinion.

• Risks are ranked by instinct, not against a stated limit or a contradicting source
• Critical and high flags arrive without the span that supports them
• The same room produces a different register depending on who ran it
• Follow-up questions are drafted without tying each to a specific risk
• By-category structure is lost in a single narrative blob

Built for senior finance teams who need committee-ready artifacts — not chat transcripts.

Risk committee

Use caseStandalone risk sweep on a new room

Moment of needThe committee wants a register it can defend line by line

OutputSeverity-ranked, cited risk register

LP underwriting lead

Use caseFirst pass before a fuller diligence

Moment of needA fast, ranked read of the risk surface before committing analyst hours

OutputRisk register and IC question pack

Allocator analyst

Use caseRe-run the register on a refreshed data room

Moment of needThe room changed and the prior register needs a delta

OutputUpdated register with by-category sections

Family office CIO

Use caseCross-cutting risk read on a manager or deal

Moment of needThe question is the risk surface, not the full recommendation

OutputExecutive summary of top risks with cited rationale

Source materials enter the engagement as a controlled register — scanned, classified, and scoped before the workflow runs.

A versioned pipeline — each stage records what happened, what you can review, and what output it produces.

Committee-ready artifacts — each with a named reviewer and a defensibility standard.

• Risk register

  One cited row per risk, ranked by severity across the eleven categories.

  Reviewed by

  Partner and risk committee

  Defensibility

  Each row cites the span it rests on — a risk that cannot be cited is omitted.

• By-category sections

  Alignment, concentration, valuation, and the rest, each readable on its own.

  Reviewed by

  Lead and partner

  Defensibility

  Rendered from the same cited rows as the register — no divergence.

• IC question pack

  One follow-up for every critical and high risk, ready to send.

  Reviewed by

  Partner before send

  Defensibility

  Each question anchors to the same source span as its risk.

Risk analysis refuses to emit a risk it cannot cite, and refuses to rank a risk critical or high without a grounded rationale. The register, the by-category sections, and the IC questions render from the same facts.

• A risk that cannot be cited is omitted, not asserted
• Critical and high require a grounded, cited rationale or the run is refused
• Deterministic ordering — the same room produces the same register
• The system supports risk judgment; it does not replace the committee

Diligence OS supports investment judgment. It does not make investment decisions, provide investment advice, or guarantee risk detection.

Prompt-based tools produce inconsistent outputs with weak provenance. This workflow runs a structured process with an admitted source envelope and a versioned review path.

Institutional outcomes — not productivity slogans.

• Standardize risk-register quality across rooms and analysts
• Surface critical and high risks early with cited rationale
• Produce IC question packs that map one-to-one to ranked risks
• Re-run a register against a refreshed room with a clear delta
• Preserve a defensible record for committee and IC sign-off
• Improve partner leverage on the risk surface

Closed by construction — each engagement runs inside a dedicated chamber with an append-only audit trail.

• Closed-system workflow — each engagement runs in an isolated chamber
• Engagement-level isolation with row-level tenant scoping
• Source-controlled review — partners read drafts against the audit log
• No training on customer documents — ever, by anyone
• Citation-backed artifacts with hash-verified source spans
• Versioned workflow history preserved on every run
• Audit trail on reads, retrievals, model calls, and locks
• Reviewer traceability — partner review recorded before sign-off

Adjacent diligence processes in the same controlled system.