The machines learned to read a contract. Now ten companies want the billable hour.

For a hundred years the law firm sold one product with a very high margin: a trained human reading carefully. The associate who checked every clause, the paralegal who pulled every precedent, the partner who signed off. All of it billed by the hour, all of it defensible because nobody else could do the reading. Then a language model learned to read a contract, and roughly ten companies noticed the same opening at once.

This is a map of that field. Not a ranking - the market is moving too fast for a leaderboard to survive the week - but a comparison of where each player is strong, who actually pays them, and the verifiable numbers underneath. The pitches all sound identical. The customer lists do not.

The Land GrabThe wedge and the platform

There are two honest ways to sell software to lawyers, and every company here has picked one. The first is the wedge: find a single painful task, do it better than a human, and hide inside a tool the lawyer already opens every morning. The second is the platform: build a new home for legal work and ask the whole firm to move in. Wedges win adoption fast. Platforms win the valuation.

Spellbook is the wedge in its purest form. It lives inside Microsoft Word, where contracts are actually written, and drafts and redlines clauses without asking anyone to learn a new interface. That narrow surface area worked: roughly 4,000 firms and in-house teams across 80 countries, a $20M Series A in early 2024, and a fresh raise reported around $50M in October 2025 to push into review. The company sits in Toronto and St. John's, Newfoundland, which tells you the talent for this is not only in San Francisco.

Legora is the platform, and it is the loudest story in the category. Founded in Stockholm in 2023 (it wore the names Judilica and Leya first), it went from around $3M in recurring revenue at the end of 2024 to roughly $100M by April 2026. In March 2026 it raised $550M at a $5.55B valuation, led by Accel. Its client list reads like a Magic Circle seating chart: Cleary Gottlieb, White & Case, Linklaters, Goodwin, Bird & Bird. When a company triples its valuation on the strength of who it sells to, the who is the product.

The pitches all sound identical. The customer lists do not.

The IncumbentsThe libraries fight back

Legal research had two names for forty years: Westlaw and Lexis. Both owned the thing the startups cannot cheaply replicate - the licensed, authoritative corpus of case law and statutes. When generative AI arrived, the interesting question was whether the libraries could ship it before the startups out-marketed them.

Thomson Reuters answered by buying instead of building. In June 2023 it agreed to acquire Casetext, maker of the GPT-4-powered assistant CoCounsel, for $650M in cash, closing that August. At the time the price looked steep for a company most lawyers had never heard of. Three years on, with a research assistant grounded in Westlaw's corpus, it reads like a bargain. TR does not break out CoCounsel's revenue; it is folded into a legal segment measured in billions.

Lexis+ AI was LexisNexis's own build, launched in October 2023, grounded strictly in Lexis content with linked citations so a lawyer can check the machine's homework. By February 2026 the company had already moved past it, replacing Lexis+ AI in the US with Lexis with Protégé, an end-to-end workflow platform. The incumbents are not standing still; they are renaming and rebuilding faster than the press can keep up.

The ReviewersContract review, four ways

The largest single job in this market is reading agreements against a set of rules. Four companies attack it from different corners of the map.

Luminance, out of Cambridge, England, grew from university mathematicians and early money from the late Mike Lynch's Invoke Capital. It specializes in due diligence and M&A review - the work that used to fill a room with associates for a week. It raised a $75M Series C in February 2025 led by Point72, and claims use across all Big Four firms and more than a quarter of the Global Top 100 law firms, with customers like AMD, BBC Studios and Hitachi.

LegalOn Technologies is the quiet giant. Founded in Tokyo in 2017 by two corporate lawyers, it builds attorney-written playbooks into its review engine and had passed 7,000 paying enterprise customers by April 2025 - more paying logos than almost anyone else on this page. A $50M round led by Goldman Sachs Growth Equity in July 2025 came with an OpenAI partnership.

Robin AI, founded in London in 2019 by a former Clifford Chance lawyer and a machine-learning scientist, pairs software review with human oversight and counts PwC and KPMG among its users. It is also the cautionary note in the set: a planned funding round of roughly $50M did not close in 2025, and the company began looking for a buyer. In a market awash in capital, a round that does not close is its own kind of signal.

LawGeex, founded in Tel Aviv in 2014, was an early pioneer of automated contract approval against a general counsel's playbook, with customers including eBay, HP and White & Case. It has also gone quietest - recent revenue, valuation and headcount are not public - a reminder that being early is not the same as staying ahead.

Notice what separates these four. Luminance sells to the deal team, LegalOn to the in-house counsel, Robin AI to the professional-services buyer, LawGeex to the enterprise procurement function. They are described in the same three words - AI contract review - and yet they barely compete for the same purchase order. In a young market, the label on the category tells you almost nothing; the identity of the person signing the invoice tells you everything.

The Workflow OwnersWhere the contract actually lives

A contract is not a document; it is a lifecycle - drafted, negotiated, signed, then referenced for years. Two companies sell that whole arc, and both existed before the AI wave gave them a second act.

Ironclad runs contract lifecycle management for names like L'Oreal, Salesforce and Mastercard. Founded in San Francisco in 2014, it raised a $150M Series E at a $3.2B valuation in early 2022 and, by January 2026, reported crossing $200M in ARR at around 40% annual growth. Its CEO, Dan Springer, previously ran DocuSign, which keeps the company's signature-and-contract lineage intact.

Clio Duo is AI folded into practice management rather than a standalone product. Clio, based in Vancouver and founded in 2008, raised $900M at a $3B valuation in July 2024 - the largest cloud legal-tech deal on record - and runs past $200M in ARR. Duo, generally available since late 2024, is the bet that solo and small-firm lawyers want AI inside the system they already use to run the business, not a separate genius tool on the side.

The ReadWhat the numbers are telling you

Three patterns fall out of this field once you line the companies up. The first: geography stopped mattering. The leaders sit in Stockholm, London, Cambridge, Tokyo, Tel Aviv, Toronto and San Francisco. A category this global this early usually means the underlying demand is real, not a single ecosystem talking to itself.

The second: the money has split into winners and orphans. Legal tech raised a record ~$6B in 2025, with 14 rounds above $100M, and Harvey alone climbed to an $11B valuation by March 2026. In the same window, Robin AI could not close $50M and LawGeex went dark on its metrics. Abundant capital does not lift every boat; it widens the gap between them.

Abundant capital does not lift every boat. It widens the gap between them.

The third, and the one that matters most to anyone billing by the hour: the work being automated is the profitable, repetitive middle - review, first drafts, precedent-pulling. None of these tools replace judgment, and none claim to. What they delete is the volume of routine reading that firms have quietly billed at a premium for a century. That is not a feature release. It is a slow repricing of a profession, and the companies on this list are the ones writing the new sticker.

There is a fourth pattern worth naming, and it is about strategy rather than money. The incumbents - Thomson Reuters, LexisNexis, Clio - are winning by owning something the startups cannot buy cheaply: the corpus, the practice system, the trust of a subscriber base that already sends them a check every month. The startups are winning by owning speed and focus. The tell for the next two years will be which side crosses into the other's territory first, and whether the acquisitions keep coming. Thomson Reuters buying Casetext, LexisNexis folding AI into Protege, Microsoft circling Robin AI - consolidation is already the loudest verb in the room.

Who wins? Ask again next quarter. The safer observation is that the question every legal AI founder is really answering - can software do a lawyer's reading? - has already been answered yes by 7,000 paying customers at one company alone. The rest is distribution, corpus, and trust. Those, it turns out, are the hard parts.