USE CASE // DEAL TEAM INFRASTRUCTURE
Beyond AI Due Diligence Software.
The first generation of AI due diligence software solved data retrieval. askOdin solves deterministic evaluation: it tests the deck against the financial model and ties every finding to a page or cell. Same documents. Different category of output. For a corporate venture (CVC) team, that record is also what headquarters reads after the people who made the decision have rotated on.
110,000+ Clarity Scores™ scored on public deal data · 4 U.S. provisional patents pending
// The Evolution
Extraction is solved. Evaluation is the new bottleneck.
The first wave of AI due diligence software did exactly what it promised: it made analysts faster. It could search a 500-page data room, summarize a CIM, and format a memo. That was a necessary first step.
But retrieving data is not the same as underwriting risk. When a partner signs a memo, they aren't verifying that the AI summarized the documents correctly. They are verifying that the business physics actually work.
That is the work the next generation of infrastructure has to do. The application layer is loud; the layer beneath it — deterministic, reproducible, audit-grade — is where capital allocation actually gets defended.
// THE PHYSICS PROBLEM
Diligence does not scale linearly. Capital deployment does.
A two-partner fund processing 2,000 decks a year cannot manually interrogate every narrative. That is not a rigor problem; it is a physics problem. When volume scales, partner judgment degrades — you start pattern-matching the logo, the warm intro, the rhyme with the last winner. And that is exactly the surface area generative AI is now flooding with prettier-sounding decks. Legacy AI due diligence software accelerates how the analyst reads, but does not change what gets read. The triage layer was already strained. It is about to break.
Inbound volume at a typical multi-stage fund.
Effective partner bandwidth for first-meeting depth.
Logo, warm intro, partner pattern-match — not physics.
// THE COMPILE PIPELINE
Four Compile Steps. One Defensible Output.
Ingest
Pitch deck, financial model, cap table, term sheet — processed as a single logic graph.
Compile
RUNE Protocol™ stress-tests every claim across five scored pillars and three audit checks against the Judgment Graph.
Triangulate
RAVEN Protocol™ cross-references documents to surface contradictions a single-document review would miss.
Log
Defensible Audit Log generated per deal — citation-grade evidence trail for IC and LP review.
The architectural mechanics of RAVEN's triangulation engine are protected under U.S. Provisional Patent No. 63/994,876 and are not publicly disclosed.
// The Distinction
Workflow accelerates. Judgment compiles.
01 · Legacy Workflow
Probabilistic LLMs that summarize what the founder wrote.
Faster reading. Tidier memos. The output is a summary — not a verdict, not an audit trail. A summary cannot be defended to an LP.
02 · askOdin Infrastructure
Deterministic protocols that compile the math against 110,000+ benchmarked public-deal cases.
Same input, different category of output: a Clarity Score, a brittle-assumption inventory, and a Defensible Audit Log™ per deal. The same recorded variables return the same score under the same version, whichever analyst ran it, and a past result is checked from its record.
03 · The Gap, Concretely
A summary can be wrong by being right.
An LLM will tell you the deck projects $10M ARR. askOdin will tell you the cap-table mathematically contradicts the projection — and flag it before the term sheet, not after the markdown.
// Corporate Venture Capital
For corporate venture teams: the questions headquarters will ask.
How should a CVC justify investment decisions to the parent company?
With the record, not with recollection. Headquarters finance and the board ask how a decision was reached, often after the team that made it has rotated. A Clarity Score comes with a Defensible Audit Log that ties every finding to the page or cell it came from, so the answer does not depend on who is still on the team.
How can a corporate venture team standardize due diligence?
By applying one standard to every deal instead of one template per analyst. Clarity scores each deal on the same five pillars, 20 points each, with three audit checks that can cap a pillar whatever it scored on its own. The CVC keeps its own sourcing and approval process; the reasoning is tested the same way every time.
Can the record support a later write-down or impairment review?
Yes, with one limit: the record shows how the reasoning was tested, not whether the investment worked. The Defensible Audit Log for each deal is a sealed, mathematically traceable record. It holds the score and its band, the version of the Clarity Framework that produced it, and every finding traced to a page or cell, tied to the SHA-256 fingerprints of the files the audit read. When headquarters considers a write-down or an impairment, finance and the auditors can read the record without re-running the audit, and see which claims in which documents were tested at the time of investment and what was left unresolved. Raw documents are held no longer than 30 days under a documented policy, so keep the documents themselves in your own records. If the fingerprint of your copy matches the record, it was the same document the audit read. "No edits after issue" is, today, a stated policy rather than a technical control.
The methodology version, and how to check a record →Does askOdin assess strategic synergy?
No. Strategic fit with the parent's business is a judgment only the corporate can make, and askOdin does not score it. What askOdin tests is whether the startup's own claims hold: the deck against the financial model, the narrative against the documents supplied with it. A low score or a PASS verdict is not a veto. It is the list of what has to be resolved. Whether to proceed for strategic reasons is for the CVC and its investment committee to decide. But a synergy case built on a thesis that does not hold is still a bad investment.
Rotated in from the business? Start with what to check in a deck and a model. Taking it through internal approval? See what corporate planning and finance will ask. Buying a company rather than a stake? The PE and M&A page covers acquisitions.
// For the New Investment Manager
New to investing? What to check first in a pitch deck and a financial model.
You do not need an investing background to do the most useful check there is: whether the numbers the deck tells agree with the numbers the model calculates. Here are five you can run by hand, with no tool at all. Every example is sample or illustrative output already published on this site.
-
01Does the same number agree in the deck and the model?
Take the deck's numbers one at a time and find the cell in the model that produces each. A number you cannot find, or one that does not compute, is your first issue.
Example: the deck claims a 74% gross margin on page 11. Model tab "Ops", cell C24, yields 41%.
-
02Has the comparison been swapped?
Check which customer segment and which period a good number comes from. A memo and its documents sometimes describe different things.
Example: the memo bounds enterprise churn at 4.2%. Model tab "Retention_Act" puts mid-market churn at 18.4%.
-
03Has an inconvenient number been left out?
Look for customers or periods excluded from a total. The exclusion is rarely the problem; the unexplained exclusion is.
Example: the expansion-revenue figure excludes churned pilot customers.
-
04Recompute the payback from the model yourself.
Divide customer acquisition cost by monthly contribution margin per customer, and compare the result with the payback the deck states.
Illustrative: the deck claims an eight-month payback. The model has acquisition cost at $18,400 against $1,100 of monthly contribution: 16.7 months.
-
05Re-run the runway from the model.
With the corrected figures, recompute how many months the cash lasts. Every contradiction ends up here.
Example: with the churn contradiction above, cash-out lands seven months early.
File each issue you find under the Clarity Framework pillar it belongs to, and every deal gets measured against the same standard.
How do I write it up for the investment committee?
Before the recommendation, answer three questions. If any answer is “we believe so”, that is not an answer; it is an opinion.
- Does the logic reconstruct?
- Do the unit economics actually support the revenue claim in the deck, or do they merely sit next to it?
- Where do the narrative and the model disagree?
- Not in general: which slide, and which cell.
- Will the record survive?
- Eighteen months on, when headquarters or the auditors ask, can you show what was checked, against which documents, before the money went in?
Then sort the findings by weight. In askOdin's output, JUDGE Protocol tiers them Critical, Major or Minor. A Minor finding gets one line in the memo. A Major finding gets a written answer from the deal lead before the vote. A Critical finding is a Kill Shot: the vote waits until the contradiction is resolved or the thesis changes.
The full method: the Pre-IC Reasoning Audit.
// For Internal Approval
For internal approval: what corporate planning and finance will want to know.
What it is
Software that reads the pitch deck and financial model a startup sends you and lists every place their numbers or claims disagree, citing the page or cell. Files are uploaded for each audit. An audit completes in under three minutes and leaves, per deal, a verdict, a 0–100 score and a record of the evidence. It does not decide whether to invest. The CVC and its investment committee do.
What it means for the company
The basis for a strategic investment stays with the company as a record, not in someone's memory. When a write-down or impairment comes up, management and the auditors can read what was checked at the time of investment and what was left open. Every deal is tested against the same standard, so the method does not change when the people do.
The risks: data, dependency, cost
- Data
- Your documents are never used to train an AI model. Raw documents are held no longer than 30 days; that ceiling is documented policy, automated enforcement of it is in progress, and until then deletion is executed on request. What persists past the window is derived (the verdict, the judgment analysis and structural data) and can include identifying details, so erasure requests reach it. askOdin runs shared, and isolation of the processing cache is still being hardened. Single-tenant deployment and a configurable processing region are planned, not shipped. SOC 2 Type I is in progress; SOC 2 Type II and ISO 27001 are planned.
- Dependency
- Extraction from the documents runs on an external model API (the Google Gemini Developer API, paid tier). The verdict itself is a deterministic calculation under fixed rules. The provider, askOdin Pte. Ltd., is a Singapore company founded in July 2025. There are no published customer case studies yet.
- Cost
- Pricing and pilot terms are not published on this page. Ask through the CVC briefing below.
Whether each control is stated policy, implemented, in progress or planned is set out on Security & data handling.
The audit layer for the last unaudited asset class.
Venture capital is the last unaudited asset class. askOdin provides the infrastructure to close the gap.
Building the case for your partners? Send them the sandbox, which runs a sample audit in the browser with nothing to upload, and this page. When they want to see it on a live deal, how a Clarity pilot runs sets out what it asks of the fund and what comes back.