Legal AI — Drafting, Review & Research
The buyer population — Offices of lawyers
Base industry report for 541110 →- Establishments · CanadaA
- 16,850
- Under 10 employeesA
- 87%
- Establishments · USA
- 166,972
- Employment · USA
- 1,078,148
- Payroll · USA
- $130.7B
Of 16,850 Canadian establishments with employees, 87% have fewer than ten — an industry of very small operators. Each of those is one potential account, before any filter for size or fit.
Entry signal — what decides who wins here
Structure decidesThe binding constraint is not executional. Being better than the incumbent does not, by itself, get you in — this one is cleared with capital, an asset, or a permission.
Come with the asset, or buy the business that owns it. Plant, fleet, land or quota decide this market before any operating skill does. Operators here are bought, not started.
Measured, not forecast: the share of US establishments opening in one year that were still active later. It counts good operators and bad ones together, which is exactly why it is the honest answer to “what are the odds”. It is for the whole sector rather than this market, and the ten-year figure comes from an older cohort because no younger one has reached ten years.
This is not a probability of success, and it is not a verdict on you. No survival probability is published per market, and inventing one would be worse than saying so. What the bar reads is how much of the outcome sits inside an operator's control: green means the hurdles are ones a better operator clears, red means the binding constraint is capital, an asset or a permission rather than execution. Someone arriving with an advantage this screen did not assume can win a market shown in red.
Companies named in this market · 10
The binding constraint — capital intensity
The most heavily funded corner of vertical software, and the money has already picked its winners. Legal AI works on the legal documents themselves — it drafts a clause, redlines a counterparty's paper against a playbook, answers a research question with citations, summarises a deposition or a data room. How it differs from the two neighbouring records: practice management (541110, Clio and peers) runs the firm's back office — matters, time, billing, trust accounting — and contract lifecycle management (541514) is the company's repository and approval workflow for executed contracts. Legal AI sits in Word and in the research database, on the work product, and is bought by the lawyer per seat rather than by finance or operations. The two neighbours are now reaching in (Clio bought vLex and its Vincent assistant for $1B; CLM suites ship AI review), which narrows the gap further. Two incumbents own the content that grounds the answers. Thomson Reuters bought Casetext and its CoCounsel assistant for $650M in 2023 [B], and its Legal Professionals segment reported $2,868M of 2025 revenue, up 8% organic, with growth credited to Westlaw, CoCounsel and Practical Law [A]. RELX's Legal segment (LexisNexis) reported £1,806M, up 9% underlying, and says over half of all US new and renewing customers are adopting Lexis+ AI [A]. Neither breaks out an AI line, but both are growing faster on a $2–3B base than most challengers are on a venture base. The challengers are funded at a scale no newcomer can match. Harvey has raised more than $1.5B, last $550M at a $15.5B valuation co-led by Diffusion and Lightspeed in September 2026, and is reported above $400M ARR with about 80% of the Am Law 100 [B; ARR and customer figures C]. Legora (Stockholm) raised a $550M Series D led by Accel in March 2026 and a $50M extension in May, $600M in that round alone at a $5.6B valuation [B]. Eve (plaintiff firms) raised $103M at a $1B valuation led by Spark Capital, $150M in total [B]; Luminance (Cambridge, UK) raised a $75M Series C led by Point72, over $115M in twelve months [B]. The Canadian name is Spellbook (Toronto): a $50M Series B led by Khosla Ventures at a $350M valuation (October 2025), over $80M of equity in total, then $40M of RBCx debt in March 2026 to buy smaller contract-AI competitors, with a two-year exclusive with the Canadian Bar Association's 40,000+ members [B; its 'on track for $100M ARR' is C]. Wordsmith (Edinburgh, in-house teams) raised a $25M Series A led by Index Ventures [B]. The middle is where companies die. Robin AI, ranked tenth on the Sunday Times 100 Tech in 2025, failed to close a $50M round and was put up for distressed sale nine months later with about $10M ARR [B]; Spellbook is explicitly raising debt to absorb competitors of that size. Capital intensity decides it: model costs, legal engineers embedded with customers and a sales motion into Am Law firms are being paid for with rounds of $500M, the content moat sits with Thomson Reuters and LexisNexis, and the sub-scale vendors are being bought or shut down rather than displacing anyone.
Sectors joined: Legal Tech · LegalTech · Legal Services · AI/LegalTech
[UNVERIFIED] Sector-to-NAICS mapping is analyst judgment — see data/angel-sector-map.json. Counts are a per-record cross-reference and are not additive across records.
The incumbent
Who owns this market and who is coming for it. Fields a screen never reached say so rather than guessing.
Financials & market size — sourced
Figures that came from a filing, a results release or reputable reporting, each carrying its evidence tier.
Market size, derived
Built from the competitor set upward rather than quoted from a forecast. Published TAMs in these categories are frequently reverse-engineered from each other, so any published figure is checked against the vendor arithmetic rather than trusted on its own.
Disclosed revenue from 1 of 9 named vendors. The market is at least this large.
No vendor has both a disclosed revenue and a published share.
Only a revenue floor is known — the true market is larger by whatever the undisclosed vendors earn.
Competitor set · 9 named · 1 disclose revenue
| Name | Revenue | Share | Note |
|---|---|---|---|
| Thomson Reuters (CoCounsel, Westlaw, Practical Law)A | $2.9B | — | Legal Professionals segment $2,868M in 2025, +8% organic (Q4 $738M, +9%) [A, 2025 results on SEC 6-K]; growth credited to Westlaw, CoCounsel and Practical Law, with 'strong double-digit growth' from CoCounsel and CoCounsel Drafting. Bought Casetext (CoCounsel) for $650M, announced June 2023 [B]. No AI line broken out |
| LexisNexis (Lexis+ AI, Protégé)A | not disclosed | — | RELX Legal segment £1,806M in 2025, +9% underlying; adjusted operating profit £415M, +12% [A, RELX Annual Report 2025]. 'Over half of all US new and renewing customers are adopting Lexis+ AI'; Protégé assistant launched Q1 2025; Lexis Create+ drafts inside Word; alliance with Harvey [A]. Revenue left null because it is reported in sterling |
| HarveyB | not disclosed | — | San Francisco. More than $1.5B raised: $160M at $8B led by a16z (Dec 2025), $200M at $11B co-led by GIC and Sequoia (Mar 2026), $550M at $15.5B co-led by Diffusion and Lightspeed (Sep 2026) [B; round confirmed in Harvey's own release]. ARR reported above $400M, 3,000+ organisations, about 80% of the Am Law 100 [C, company via press] |
| LegoraB | not disclosed | — | Stockholm (formerly Leya). $550M Series D led by Accel at $5.55B (Mar 2026) with Benchmark, Bessemer, General Catalyst, ICONIQ, Redpoint, YC and others [A, company release], plus a $50M extension (Atlassian, NVentures) in May 2026 — $600M in the round at $5.6B [B]. ARR $100M+ and 1,000+ organisations claimed [C]. Earlier rounds not opened for this record |
| SpellbookB | not disclosed | — | Toronto — the Canadian challenger. Contract drafting and review inside Word. $50M Series B led by Khosla Ventures at a $350M post-money valuation (Oct 2025), over $80M equity in total; $40M RBCx debt (Mar 2026) to acquire contract-AI competitors; exclusive two-year partnership with the Canadian Bar Association (40,000+ members) [B, BetaKit]. 'On track for $100M ARR in 2026', 4,000+ customers in 80 countries [C] |
| LuminanceB | not disclosed | — | Cambridge, UK. $75M Series C led by Point72 Private Investments (Feb 2025), over $115M raised in twelve months; earlier a $10M round at a $100M valuation (2019) [B]. 700+ organisations in 70+ countries claimed [C]. Sells to corporate legal as much as to firms, so overlaps CLM |
| EveB | not disclosed | — | San Francisco; AI for plaintiff-side firms (intake, records, demand letters, discovery). $103M Series B led by Spark Capital at $1B (Sep 2025), $150M total, with a16z, Lightspeed, Menlo [B]. 450+ US firms claimed [C] |
| WordsmithB | not disclosed | — | Edinburgh; AI for in-house legal teams (contract and policy review, a Word plug-in, an internal legal help desk). $25M Series A led by Index Ventures with Scottish Enterprise, valuation above $100M (Jun 2025) [B] |
| Robin AIB | not disclosed | — | London; AI contract review. Failed to close a $50M round and was listed for distressed sale in October 2025 with about $10M ARR [B, Legal IT Insider]; its managed-services arm and engineers were reportedly dispersed after the sale [C, not opened]. The cautionary case for a sub-scale entrant |
Startups & challengers
Newer and smaller vendors going at the incumbent — funded challengers first. Named, not researched to the depth of the field above; a company with a page here links to it.
| Company | Stage | What it does | Raised |
|---|---|---|---|
| Disco | Funded challenger | Cloud ediscovery and AI document review for law firms and legal departments | — |
Evidence
Evidence. Incumbent scale is tier A: Thomson Reuters' fourth-quarter and full-year 2025 results (SEC 6-K exhibit 99.1, opened) give Legal Professionals at $2,868M, +8% organic, and RELX's Annual Report 2025 (filed on 6-K, opened) gives Legal at £1,806M, +9% underlying, with the Lexis+ AI adoption statement and the Harvey alliance. Neither company reports an AI or CoCounsel line, so the size of legal AI as a category is UNVERIFIED and no market-research total is used. Funding rounds were opened in the companies' own releases or trade press [B]: Harvey's September 2026 release (harvey.ai) and Latham & Watkins' deal note; the $11B round in Bloomberg Law; the $8B round in Pulse 2.0; the 'more than $1.5B total' and '$400M ARR' figures in Crypto Briefing's coverage of the September round (secondary; ARR is a company figure, C). Legora's Series D from legora.com's newsroom and the extension from CFOtech. Spellbook's Series B and RBCx debt from BetaKit. Luminance's Series C and 2019 round from Legal IT Insider (Global Legal Post and Luminance's own press page returned 403/404). Eve from Legal IT Insider. Wordsmith from Tech.eu. Robin AI's distressed sale from Legal IT Insider. Casetext's $650M from the ABA Journal (Thomson Reuters' own Legal Current page now redirects). Every customer count and ARR figure in this record is the vendor's own claim [C]. Not researched: EvenUp, Definely, Irys, Genie AI and other smaller entrants; Clio's own AI is on the 541110 practice-management record.
Where the industry talks
The associations, forums and events where people in this trade actually talk shop — where to listen before entering, and where the first customers are found. Each link was opened on the date shown.
National association of Canadian lawyers with provincial branches and an in-house lawyers section; in March 2026 named Spellbook its exclusive AI contract drafting and review partner for its 40,000+ members (BetaKit) — the channel is already taken
Peer association of law-firm and legal-department technology staff — the people who evaluate AI tools; runs ILTACON, ILTACON Europe and a Legal Tech On Tour Canada stop. No member count on home page
Community for legal operations professionals in corporate legal departments — the in-house buyer for Wordsmith, Luminance and Harvey's corporate tier; runs a Global Institute and an AI Intensive
American Bar Association Law Practice Division's legal-technology conference and expo; next March 3–6, 2027, McCormick Place, Chicago
London legal-tech conference (14–15 October 2026, Old Truman Brewery) with North America and Europe editions and a startup map
Legal tech and AI news site; covers every funding round in this record
Legal IT news; source for the Legora, Luminance, Eve and Robin AI items here
Robert Ambrogi's legal-technology blog and the LawNext podcast
Association of Corporate Counsel (acc.com) and the Law Society of Ontario (lso.ca) returned bot checks and are not listed. Legalweek (legalweek.com) redirects to Law.com and is not listed.
The businesses it sells to
Operating-business records filed along the same branch of NAICS — the customers of this software, screened as businesses in their own right.
No operating-business record has been written along this branch yet. The base industry report says what the subsector typically runs on.
Other software on this branch
Immigration case-management software fills government forms from one client questionnaire, tracks receipts, priority dates and expiries, and runs a multilingual client portal. The buyer is the immigration practitioner: immigration law firms first, but the same products are sold to regulated immigration consultants (Canada's RCICs, licensed by the College of Immigration and Citizenship Consultants), to non-profits, and to corporate mobility and HR teams that sponsor visas. It differs from the neighbouring records in that general practice management (541110 Legal Practice Management Software — Clio, MyCase) is horizontal across practice areas, legal AI drafting is a document tool, 9113 is the government side of the filing, and the employer-of-record record covers hiring abroad without a visa. The practitioner tier is already owned by the legal-software consolidators. In one year, 2022, three of the best-known independents were bought: Paradigm (PracticePanther, Bill4Time, MerusCase) bought LollyLaw in April [B: LawNext]; MyCase bought Docketwise in May [B: ABA Journal], and AffiniPay bought MyCase the same year — the group rebranded as 8am in August 2025 and says it serves 250,000+ professionals across the US and Canada [B: LawNext; customer count C]; and Equifax bought LawLogix (Guardian for I-9, Edge for case management) from Hyland, closing August 2022, terms undisclosed [A: Equifax release]. Mitratech had already bought INSZoom in November 2020 and calls itself 'the global leader in immigration case management software' [C, vendor release]; INSZoom now sells U.S., Canadian and global case processing. Around them sit the long-lived independents — Cerenade's eimmigration (Los Angeles, founded 1992 [C]) and CampLegal (Philadelphia, founded 2017, 3,800+ immigration professionals claimed, from $69 a month [C]). Canada's RCICs are served by Officio (Uniques Software) and by INSZoom's Canadian module, plus a crowd of very young RCIC-only tools that do not yet meet this record's bar. The corporate side is being taken by firms that are software and law firm at once. Envoy Global (Chicago; Palladium Equity Partners since September 2021, True Wind Capital and Elyan Partners in January 2022 [B]) sells software with in-house legal teams and has bought practices in London, Dubai and Cape Town. Boundless ($45.3M raised to April 2021 [A, own release]) now advertises through its own Arizona-licensed law firm. Alma, 'a modern immigration law firm', raised a $17M Series A led by Bling Capital in September 2026 on top of $5.1M seed [B]. These compete with the customer, not with Docketwise — which squeezes the law-firm buyer from above. Incumbent vulnerability decides it. Every seat a new vendor would want has an owner with a payments or HR-data business behind it, the form library has to be re-certified every time USCIS or IRCC revises a form (INSZoom promises updates within 24–48 hours), and the products are cheap enough that price is no wedge. One real weakness: Docketwise disclosed that credentials to third-party partner repositories were used to clone data in October 2025, including passport and Social Security numbers of law firms' clients; it began notifying individuals on 3 April 2026 [C, vendor notice, opened]. Trust is a selling point here, but a single incident at one brand does not open a market to an unfunded entrant. The only angle worth keeping is Canadian: an RCIC-first product built by someone inside a consultancy, sold through CAPIC — and even there INSZoom and Officio are already in the room.
Clio just raised $500M at a $5B valuation and spent $1B buying vLex. A market whose leader is actively bundling research, drafting and payments into one suite is the worst possible moment to arrive with a point solution.