What it is
Crosby is a law firm, not a software product. It reviews and negotiates commercial contracts — NDAs, MSAs, DPAs and procurement agreements — with AI agents doing the first pass and licensed attorneys signing off on what goes back to the client. The legal work is delivered by Crosby Legal PLLC; the agents are built by Crosby Legal, Inc. The firm states it carries full malpractice insurance, and its Series B announcement says “we stand behind our work and take liability for it.” No review tool on the market makes that promise.
Ryan Daniels (CEO, formerly at Cooley and a startup general counsel) and John Sarihan (CTO, an early Ramp engineer) founded it in New York in September 2024, soft-launched in January 2025 and went public in June 2025. It has raised about $85.8M: a $5.8M seed led by Sequoia, a $20M Series A in October 2025 with Index, Bain Capital Ventures and Elad Gil, and a $60M Series B on 31 March 2026 led by Lux and Index, at a reported $400M valuation. The site now claims 50+ licensed attorneys. Named clients include Cursor, Clay, Cognition and Tishman Speyer; Forbes reported Cursor had sent about 2,000 contracts through Crosby by March 2026. On 29 October 2026 Crosby filed a letter of intent for the brand and some internal data of the winding-down firm Weil, Gotshal & Manges; terms were not disclosed and no outcome has been reported.
Why it shows up in Legal Ops stacks
- It replaces the work, not the lawyer’s tooling. You send a contract through Slack or email; it comes back reviewed against a playbook Crosby keeps for your company, with redlines and the negotiation position. Nobody on your side has to learn a Word add-in, which is the whole point for a company with no lawyer to drive one.
- Turnaround is measured in hours. Crosby reported a 58-minute median review time at launch in June 2025, and in October 2025 said it still returns over 90% of documents within a few hours. That is the bar outside counsel usually misses on a Friday-afternoon order form.
- The price is per outcome. One fee per contract by document type, not hourly billing, and third-party coverage reports one charge covers repeated review rounds on the same agreement.
Pricing reality
Crosby publishes no price list; “Learn about pricing” on its site leads to a sales form. Forbes reported $250–1,000 per contract in March 2026, roughly $10–50 per page. Treat the bottom of that band as an NDA price and the top as a negotiated MSA or DPA.
The math decides the purchase. At a $400 midpoint, 15 contracts a month costs about $72,000 a year. A GC AI seat is published at $500 a month, about $6,000 a year — the same as roughly one Crosby MSA per month. Software wins as soon as you have a lawyer who will use it; Crosby wins when the alternative is a founder, a sales leader or a lone GC reviewing paper themselves, or outside counsel at hourly rates. Axiom reports a 2026 average of $152 an hour for its commercial-contracts work, so a three-hour MSA review through an ALSP lands mid-band at about $456 before any second round.
Best for
A COO, head of sales ops or solo general counsel at a venture-backed US company doing roughly 10–60 inbound customer and vendor contracts a month on standard templates, where deal speed matters and nobody in-house has time to review. That is the Cursor and Clay profile, and it is where Crosby’s per-contract fee undercuts both a new legal hire and outside counsel.
Skip it if you already have three or more commercial lawyers in-house: at a few hundred contracts a month, the per-document bill runs into six figures a year and a review tool your team drives is far cheaper. Skip it too for litigation, M&A, regulatory work or bespoke deals — Crosby sells commercial contract review, and nothing on its site covers anything else.
Versus the alternatives
On the “buy a tool and keep the work in-house” side, Spellbook is the volume leader in Word-native contract AI; pick it when a lawyer drafts and redlines all day. LegalOn is the playbook-driven review platform built for in-house teams; pick it when a legal team of three or more wants attorney-written playbooks applied consistently. Ivo competes for the same in-house review buyer, and GC AI is the self-serve route with a published price for a single in-house lawyer.
On the “buy the labor” side, Axiom is the established ALSP; pick it when you need a seconded lawyer embedded in your team rather than document-by-document review. The fastest-growing comparable model is Eudia, which pairs software with acquired ALSPs and its own law firm, Eudia Counsel, priced as a fixed annual fee — pick it at Fortune 2000 scale when you want to convert an entire outside-counsel budget, not a contract queue. Pick Crosby when volume is moderate, templates are standard, and nobody in-house can own review. If none fit, keep outside counsel for bespoke paper and put an intake tool such as Checkbox or Sandstone in front of it so only the contracts that need a lawyer reach one.
Watch-outs
- Per-contract pricing scales linearly with deal volume. A sales team that doubles its pipeline doubles the legal bill. Guard: get the rate card by document type in writing, model next year’s contract volume, and set the threshold — a review-tool seat costs about as much as one or two Crosby contracts a month — at which a hire plus a review tool becomes cheaper.
- Your playbook lives with Crosby. Positions are captured in Crosby’s per-client memory, not in a document you own. Guard: write your fallback positions down yourself, and require in the retainer that Crosby exports the playbook and the redline history of every agreement on request.
- Admissions and data terms are not on the public site. Crosby names no bar admissions, its readable pages state no SOC 2 attestation or model-training policy (the trust center needs access), and its terms-of-service page is still a September 2024 template saying the service “does not provide legal advice.” Guard: before sending customer DPAs, get the retainer naming the responsible attorneys and covered governing law, the malpractice limits, and the security pack, including whether client contracts train the agents.