A 14-page term sheet, graded clause by clause, in ninety seconds.
Upload the PDF. Twelve standard clauses are extracted deterministically, scored founder-friendly through hostile, and paired with a suggested counter. No AI guessing. No black box.
No card · Your document never trains a model · Results in ~90 seconds
Term sheets are a foreign language designed to sound boring.
Most founders see their first one, skim for the valuation, and sign. The real damage is in the thirteen pages of defined terms — liquidation preference, anti-dilution, drag-along, pay-to-play. Ignore them and you can take home a fraction of what the headline number promised at exit.
Length of a typical Series A term sheet. One of them is a valuation. Thirteen determine payout.
How much a 2× participating preference can cost founders at a mid-size exit vs 1× non-participating.
How long it takes Deal Intelligence to extract, grade, and counter every standard clause.
Three moves. Deterministic every time.
Twelve clauses, pulled verbatim from the PDF.
The extractor is deterministic, not a language model. It locates liquidation preference, anti-dilution, board composition, drag-along, pro rata, redemption rights, option pool, vesting, protective provisions, information rights, pay-to-play, and conversion. Every finding carries the page and paragraph it came from.
Each clause scored founder-friendly to hostile.
The verdict engine weights each clause against current market norms. 1× non-participating preferred with broad-based weighted average is standard; 2× participating with full ratchet is not. Scores are numeric (0–100 per clause) and the weights used are visible — not hidden behind a prompt.
Every hostile clause gets a suggested counter — with the dollar swing.
When a clause scores hostile, the engine proposes a specific counter (e.g. 1× non-participating instead of 2× participating) and runs the exit waterfall at $10M, $50M, $200M, $1B so you see what you gain back. You walk into the call with a number, not an argument.
You know exactly what you're signing before the call with the lead.
After one upload:
A verdict in plain English.
Every clause restated in a sentence, with the score and what it means for your take at exit.
A counter-offer for every hostile clause.
Specific asks, grounded in market norms, with the dollar swing at four exit sizes.
A shareable analysis.
Export the verdict as a PDF your lawyer, co-founder, or advisor can read in five minutes.
No AI vendor lock-in.
The extractor is deterministic. You could audit it if you wanted to. Your document doesn't train anyone's model.