Two engines, both deterministic. A rules-based kill-list produces your score and gap list; a convertible-instrument model computes your dilution from your own cap table. No opinion, no black box. The arithmetic below is exactly what runs.
The kill-list is drawn from general capital-markets diligence practice — the same categories that appear in public model documents (NVCA), accelerator resources (YC's SAFE materials), and law-firm data-room lists (e.g. Cooley GO). Every line is written fresh in plain English. It contains no jurisdiction-specific regulatory language and no employer-confidential material.
Items span seven categories:
Each item carries a weight: 3 = a deal-killer or re-pricer if found late, 2 = material cleanup, 1 = hygiene. Each item lists the rounds where it's expected (a Series A diligence checks more than a pre-seed). Your score:
The LLM writes nothing here. The score, the gap list, and the ordering are pure rules — the same code runs client-side on the homepage and server-side in your report, so the grades match.
The engine takes your existing fully-diluted shares, your convertible stack (SAFEs / notes with their terms), and a priced round (new money, pre-money valuation, optional pool refresh), and solves the post-round cap table by fixed point. The finance, from first principles:
Each SAFE converts at whichever term gives the holder the most shares — i.e. the lowest price per share among:
investment ÷ cap of the company measured on the pre-new-money capitalization — which, per the standard definition, includes the promised option pool. So post-money SAFEs do not dilute each other; they dilute together with founders when the new money comes in.A bare SAFE with an MFN right and no cap/discount of its own inherits the best terms granted to any other SAFE — the lowest cap and the highest discount on offer — since "best" means whatever yields the lowest conversion price. In the sample report, the MFN angel inherits the $6M cap and converts identically to the capped seed SAFE.
When investors require a target post-round option pool, it is created pre-money. Because it's carved before the new money, its entire dilution lands on existing holders (founders + converting SAFEs), not the incoming investor. The report shows the points of ownership this quietly moves:
For holders with a pro-rata side letter, the report shows the dollars they'd need to invest in a same-sized next round to hold their converted stake — the industry framing ownership% × next-round size — so you can see the follow-on reserve your existing investors can claim.
dilution.test.ts) that hand-derives 2–4 scenarios from first principles and asserts the code matches — a cap-bound post-money SAFE, a discount-beats-cap case, an MFN + pool-shuffle case, and the pool-shuffle helper. It is arithmetic, not a valuation, and it only reflects the inputs you give it.