Venture capital is the last unaudited asset class. Credit has Moody's. Public markets have GAAP. Private capital has gut feel — three reference calls, one warm intro, and a partner remembering when this kind of deal worked the last time.
That has been good enough for forty years. It is not going to be good enough for the next ten. The market does not need another AI that writes prettier summaries. It needs a verification layer.
askOdin is that layer. We are the AI Judgment Infrastructure for private capital.
Every Institutional Function Already Has One
This is not a novel idea. It is the last unbuilt instance of a very old one. Each time a decision became too consequential to leave to individual discretion, the industry around it built a standard, and then built infrastructure to enforce the standard.
| Institutional function | Governing standard | What it makes checkable |
|---|---|---|
| Financial reporting | GAAP · IFRS | Whether the books say what they appear to say |
| Audit | GAAS · PCAOB | Whether the examination itself was rigorous |
| Information security | SOC 2 · ISO 27001 | Whether a control is implemented or merely written down |
| Investment judgment | AI Judgment Infrastructure™ | Whether the reasoning behind an allocation holds |
No enterprise closes its books on an unstandardised ledger. No auditor signs an opinion on a method they invented that morning. But an investment committee will approve an eight-figure allocation on a number somebody copied off a slide, and nobody in the room can say who checked it. That is the last gap on the list.
Closing the Audit Gap
The Audit Gap is the void between a founder's narrative persuasion and the structural reality of their business. Historically, identifying this gap required weeks of manual interrogation. askOdin collapses this process into a deterministic compilation sequence, identifying structural contradictions before capital is deployed.
The Benchmark Corpus: 110,000+ Clarity Scores on Public Deal Data
To evaluate judgment, you need a baseline of business physics. askOdin operates the Judgment Graph™ — a benchmark corpus of 110,000+ Clarity Scores built on public deal data.
- The Corpus
- 110,000+ documents of public deal data, scored.
- The Basis
- A benchmark corpus of 110,000+ Clarity Scores built on public deal data.
- The Advantage
- No customer document trains anything here. The engine is a deterministic compiler executing a fixed rule set; the 110,000+ Clarity Scores in the benchmark corpus were produced BY that engine on public deal data, not used to tune it.
Moat Specification
// THE INFRASTRUCTURE STACK
The Architecture
askOdin is not a software application; it is an infrastructure stack engineered to issue institutional-grade judgment. The RUNE Protocol™ is the deterministic judgment compiler at its core.
What Actually Gets Scored
Five pillars, twenty points each. Three audit checks — business physics, unit economics, market evidence — that can cap any pillar regardless of what it scored on its own. Design, wording and delivery are worth nothing. The rubric does not vary by firm, by sector, or by who is running it; that invariance is the entire point of calling it infrastructure.
CLAIM 01
Problem Definition
Is the pain structural or cosmetic? Severity, urgency, switching inertia, and whether a regulatory mandate is forcing the purchase or a founder is hoping for one.
CLAIM 02
Solution Logic
Does the mechanism survive its own physics? Technical feasibility, single-point dependencies, and whether the solution violates a constraint the market cannot lift.
CLAIM 03
Market Evidence
Is demand pulling, or is supply being pushed? TAM arithmetic, competitive density, and pricing power that holds under contact with a real buyer.
CLAIM 04
Business Model Physics
Do the unit economics scale, or compound losses under load? True gross margin, CAC and LTV decay, burn multiple, working-capital drag. This is where two decks in three lose the most points.
CLAIM 05
Deal Structure
The ask and the milestones. Is the raise sized to what it has to prove, and does the cap table survive the next round? Shown as The Ask inside the product.
A contradiction inside any pillar is flagged as a Compile-Time Error — surfaced before a single token of analysis is generated, not discovered afterwards in a summary that already reads as confident.
Four Generations of Software in Private Capital
Retrieval retrieves. Analytics aggregates. Workflow automates. Each generation solved a real problem and left the consequential one untouched: none of them reads the reasoning. We do not compete with these systems; we consume their output.
| Generation | Mechanism | What it cannot do |
|---|---|---|
| Market intelligence PitchBook, AlphaSense | Indexed retrieval — who funded what, and what was written about it | Tells you the deal exists. Says nothing about whether it holds. |
| Relationship workflow Affinity and peers | Entity graph over people and pipeline | Tracks who knows whom. Never opens the memo. |
| Generative assistants ChatGPT, Copilot, RAG tools | Probabilistic token generation over retrieved context | Summarises the argument fluently, including the parts that are wrong. A contradiction between two documents is smoothed into a plausible reconciliation, because that is what self-attention is built to do. |
| Judgment Infrastructure™ askOdin | Deterministic compilation into a logic-validated dependency graph, evaluated outside the model | Does not predict outcomes, and does not pretend to. It measures whether the reasoning justified the cheque. |
A language model is optimised to be agreeable. Hand it a deck claiming 140% of a serviceable market and it will render that claim in clean prose. The RUNE Protocol™ treats it as a variable with a constraint, finds the constraint violated, and cascades the failure through everything downstream that depended on it. LLMs optimize for persuasion. askOdin compiles for physics.
The Same Contrast, Attribute by Attribute
The generational table above maps the market. This one drills into the single boundary that matters — what separates probabilistic text generation from deterministic judgment, line by line.
| Attribute | Legacy Generative AI | AI Judgment Infrastructure (askOdin) |
|---|---|---|
| Core Mechanism | Probabilistic Summarization | Deterministic Compilation |
| Primary Output | Text Generation & Slide Formatting | Defensible Audit Log & Clarity Score |
| Analytical Focus | Narrative Polish & Spelling | Business Physics & Compile-Time Errors |
| Market Position | Assistant / Copilot | Rating Agency / Infrastructure |
The Defensible Audit Log™
Institutional capital requires reconstructible decisions. Every Clarity Score issued by askOdin generates a Defensible Audit Log — a sealed, mathematically traceable record of exactly why an assumption was flagged or a thesis was validated.
For the founder, it is a crucible that hardens the business model. For the allocator, it is the standard of care required for modern fiduciary duty.
> RUNE Protocol — compile thesis CLARITY SCORE ........ 0–100, deterministic DIMENSIONS AUDITED ... 40+ CALIBRATED AGAINST ... 110,000+ public-deal Clarity Scores OUTPUT ............... Defensible Audit Log — mathematically traceable
Frequently Asked Questions
What is AI Judgment Infrastructure?
How is AI Judgment Infrastructure different from legacy generative AI?
What is the Judgment Graph?
What is the RUNE Protocol?
What is the Defensible Audit Log?
Deploy the standard
Intuition is personal. Judgment has to be institutional.
The category is only useful as something you can run. Two products operate on this standard, and the same score means the same thing in both.
One standard of care across the pipeline, and a record that survives an LP post-mortem.
Your own deck and model through the same rubric, in under three minutes. Free.