The customer already wants to buy. The money is stuck in the contract.
That is one of the most familiar growth frictions in B2B companies, and one of the hardest to admit publicly. Sales has the buyer’s intent. The parties are already negotiating an MSA, DPA, or NDA. The next step should be signature, invoice, and onboarding. Instead, the contract enters a legal queue, outside counsel bills by the hour, emails move back and forth, and two weeks disappear while the business team keeps asking for an update.
Legal work cannot be careless. The problem is that, inside many high-growth companies, the true delay is not only legal judgment. It is intake, clause extraction, precedent lookup, tolerance calibration, redline preparation, and response drafting.
Crosby attacks that opening.
It is not an AI plug-in sold to lawyers. It is an AI-native law firm. TechCrunch reported in June 2025 that Crosby launched with a $5.8 million seed round and a model in which AI software and human lawyers jointly deliver contract review for high-growth customers including Cursor, Clay, and UnifyGTM. At launch, the company had processed more than 1,000 MSAs, DPAs, and NDAs and promised new customer contract reviews within one hour.
The later speed is more revealing. On March 31, 2026 Crosby announced a $60 million Series B led by Lux and Index. Forbes reported the same day that Crosby had about 100 customers, had reviewed 13,000 contracts, had grown revenue about 400% since October 2025, and was valued around $400 million. Those contract volume, revenue growth, and valuation figures likely depend at least partly on company disclosures or informed sources, not a third-party audit.
Still, they point to a useful signal: the non-consensus opportunity in legal AI may not be making lawyers faster. It may be rebuilding the law-firm business as a product.
It sells an outcome, not a tool
The default legal software path is to embed AI into the lawyer’s existing workflow. The product summarizes contracts, drafts clauses, searches precedent, or suggests redlines. The customer buys software. Judgment, responsibility, and delivery stay with the lawyer.
Crosby reverses the sequence.
The customer sends a contract through Slack, email, or an existing workflow. Crosby’s system performs clause extraction, risk identification, market-term comparison, and customer-preference retrieval. Licensed lawyers then review, adjust, and issue the redlines and negotiation guidance. The customer does not receive “the AI suggests this.” The customer receives “this contract has been reviewed and can move to the next negotiation step.”
That distinction matters.
In high-risk vertical work, users often do not want a smarter tool. They want a completed result with responsibility attached. A startup CEO or sales leader does not want to learn how to prompt a legal model or decide whether a limitation-of-liability clause is reliable. The practical question is simpler: can we sign this, which terms must change, which concessions are acceptable, and when can the redline go back?
Crosby hides AI behind the service and keeps responsibility inside the delivery.
The billing unit changes the business
The traditional law-firm billing unit is the hour. Hourly billing creates a structural conflict: the customer wants speed, but speed can reduce the service provider’s revenue. The faster the lawyer works, the shorter the bill. The slower the process, the more painful it becomes for the customer.
Crosby appears to move toward fixed-price review by file or matter. Sacra’s company profile describes Crosby as using fixed per-file pricing, with a typical contract review around $400. Yespress also describes fixed per-document pricing in the several-hundred-to-one-thousand-dollar range. These are third-party profiles rather than an official public Crosby price list, so they should be treated as estimates.
The commercial logic is clear anyway. The customer is not buying the number of lawyer hours. The customer is buying a predictable contract-review result.
That changes who captures the value of AI efficiency. Under hourly billing, AI that reduces labor time can compress revenue. Under fixed per-matter pricing, AI that lets the same team process more contracts, deliver faster, and preserve quality turns directly into gross margin.
This is why Crosby is worth studying. It does not treat AI as a feature. It places AI underneath the business model. “Time” becomes “outcome.” “Human queue” becomes “system throughput.”
Why customers buy it
Contract review blocks revenue, not just the legal department.
Crosby’s Cursor customer case makes that point clearly. Cursor needed sales contracts to move in hours or minutes, not days or weeks. Crosby embedded into Cursor’s GTM workflow, letting sales teams submit redlines in Slack and receive guidance or revised documents within hours while the legal team retained visibility and control.
The case reports a roughly 50% reduction in review time, 30% growth in contract volume, and more than 2,100 requests handled. Again, these are Crosby customer-case metrics, not independent audit evidence. But they explain why the buyer is willing to pay.
If Crosby is understood as “cheaper lawyers,” the case is being underestimated.
The real budget is sales velocity. For a high-growth B2B company, a standard contract approved one day earlier can mean a customer starts sooner, revenue is recognized sooner, and sales stops chasing legal for status. The value of contract review is not only reduced outside-counsel spend. It is less transaction friction.
That is also why Crosby’s entry point is not a complicated legal management console. It fits into Slack, email, and the workflows the customer already uses. It behaves more like another team member than another system.
The product is a learning law firm
Crosby’s deeper productization is that each review can become reusable organizational memory.
In the Cursor case, Crosby describes knowledge infrastructure that records customer preferences, historical negotiations, counterparty positions, internal discussion, and strategic priorities. If legal accepts a clause in one negotiation, that choice becomes context for similar customers and similar contracts in the future.
That may sound like a knowledge base, but it sits closer to business decision-making than ordinary document storage.
The hard part of contract review is not merely whether a clause is legally valid. It is whether this company should concede this point for this customer, this deal size, this market position, and this sales stage. That judgment used to live in lawyer memory, email threads, and scattered files. New teammates could not inherit it easily, and sales could not call it up in real time.
If Crosby can turn those preferences into a queryable, reusable system that grows with the customer, it is no longer selling one review. It is selling the customer’s own commercial legal operating system.
That is the likely lock-in: not the model itself, but customer history, risk tolerance, negotiation context, and accumulated precedent.
The risks are real
This model is not light.
First, legal services have regulatory boundaries. Crosby emphasizes that it is a vertically integrated AI-native law firm with lawyers behind the work and responsibility attached. That is a selling point, but also a constraint. State-by-state, country-by-country, and matter-type expansion will be shaped by licensing, malpractice insurance, and regulatory structure.
Second, fixed pricing must survive complexity. Standard NDAs, DPAs, and MSAs can be productized. Highly bespoke, cross-border, M&A, financial, or regulated contracts may require much more human judgment. How much work AI and process standardization can absorb will determine margin.
Third, large law firms, CLM vendors, and legal AI companies will not ignore the signal. Harvey, Spellbook, Ironclad, Legora, and others are all competing for legal AI entry points. Crosby is different because it delivers the legal service itself. But if fixed-result delivery proves valuable, more hybrid service companies will follow.
So Crosby is not a simple “AI replaces lawyers” story. The more realistic question is whether professional services can be redesigned with software-like throughput while preserving expert responsibility.
What builders should learn
Crosby matters because it points to an underused AI commercialization path.
Many AI products still sell faster generation: faster copy, faster code, faster contract reading. But customers often pay the highest prices not for generation speed, but for reliable outcomes, clear responsibility, and controlled delivery.
Crosby decomposes contract review into a result product. The input is a contract. The output is lawyer-backed redlines and negotiation advice. AI compresses repeat work. A knowledge system captures customer preference. Fixed pricing replaces hourly billing.
The lesson is direct: when an industry’s old billing unit does not match the customer’s real value, AI may be most powerful when it rebuilds the billing unit.
When legal work is no longer sold by the hour, but by whether the contract keeps moving, AI enters the commercial deep water.