CoCounsel
AI legal assistant inside the Thomson Reuters Legal Professionals segment — the live successor to the Casetext acquisition. Connects Westlaw primary law, Practical Law how-to content, KeyCite citator, ONESOURCE+ tax and CoCounsel Tax/Audit through an agentic workflow rebuilt on Anthropic’s Claude Agent SDK.
The Business
CoCounsel is Thomson Reuters’ AI legal assistant — the live successor to the Casetext acquisition (closed 17 August 2023 for $650M cash). It sits inside the Thomson Reuters Legal Professionals segment and powers GenAI features across Westlaw primary-law research, Practical Law how-to content, the KeyCite citator graph, CoCounsel Tax and Audit, and ONESOURCE+. The product was first launched as Casetext CoCounsel in March 2023 and has been progressively rebuilt under TR ownership; the November 2025 announcement re-anchored the agentic platform on Anthropic’s Claude Agent SDK, with US beta in early 2026 and broader availability across 2026. Headline use cases span legal research, document drafting and review, deposition preparation, contract analysis at bulk-document scale (up to 10,000 documents), and the workflow builder where lawyers describe matters in plain language and CoCounsel plans the inquiry, retrieves citation-grounded references from Westlaw / Practical Law / KeyCite, and drafts with validated citations.
Customers and Distribution
CoCounsel passed 1 million professional users across 107 countries in February 2026 — a ~10x increase off the post-Casetext baseline of ~10,000 customer organisations, three years after the March 2023 launch. Customers span AmLaw 100 firms (Polsinelli’s firmwide rollout to ~1,200 attorneys in July 2025 is the named exemplar, with 70% associate and 50% shareholder active usage), corporate legal teams (Microsoft was cited on the TR Q4 2025 earnings call as a notable enterprise win), midsize firms, government legal, and tax / audit firms via CoCounsel Tax and Audit. Distribution sits inside Thomson Reuters’ existing direct-sales motion — one of the most mature enterprise sales organisations in legal and professional services, with decades of relationship history across AmLaw firms, corporate legal and government — and the CoCounsel product rides this motion rather than building a parallel one. The TR Legal Professionals segment recorded 97% recurring revenue with broad customer distribution and no top-5 buyer concentration. GenAI-enabled share of TR annualised contract value reached 28% in Q4 2025, up from 15% when first disclosed in Q3 2024 — nearly doubling in five quarters — and TR has stated a target to grow CoCounsel users tenfold in 2026.
Model Strategy
CoCounsel’s distinguishing asset is access to Thomson Reuters’ authoritative legal corpus: Westlaw primary law plus secondary sources, Practical Law’s curated how-to content, the KeyCite citator graph, and Reuters editorial. This is a multi-decade proprietary editorial layer that competing LLMs cannot replicate from the open web — the data moat is the defining asset that distinguishes CoCounsel from Harvey-on-public-corpus alternatives, and was named in IM’s #IM106 sector briefing as one of the clearest structural assets in legal AI. The foundation-model layer originally used OpenAI GPT-4 (Casetext era) and is now rebuilt on Anthropic’s Claude Agent SDK as announced November 2025, with the May 2026 deepening adding an MCP integration that connects Claude directly into CoCounsel Legal. Multi-cloud at the infrastructure level (TR’s broader stack spans AWS and Azure historically); single-frontier-model at the agentic-platform layer is the deliberate trade-off — supplier-diversity score sits at 6 on this evidence. Above the foundation layer, the 2026 reimagined release introduced a customisable workflow builder where lawyers can save and share workflow plans across practice groups — an emerging in-segment network-effect surface that the IM Framework scores under D5 platform dynamics for the segment, not for the parent TR network.
Leadership Team
CoCounsel is segment-led inside Thomson Reuters’ Legal Professionals reporting segment rather than a standalone subsidiary — the v1.5-A8 segment rule applies. Bench depth is via TR’s existing org (Westlaw / Practical Law / KeyCite product management, TR Labs research, TR enterprise sales), which reduces key-person fragility well below typical AI-startup levels. Other senior TR executives publicly active on the CoCounsel thesis include CFO Mike Eastwood and Chief Product Officer David Wong.
IM Framework Scoring
IM’s structured assessment of CoCounsel’s competitive position. The summary below is the headline; expand “Show the full analyst-grade analysis” near the bottom for the per-dimension reasoning and evidence. Methodology →
Funding History
| Date | Round | Raised | Post-money | Lead investor(s) |
|---|---|---|---|---|
| Aug 2023 | Acquisition by Thomson Reuters (closed) | $650M cash | — | Thomson Reuters (NYSE: TRI) |
| Mar 2023 | CoCounsel launch (Casetext, pre-acquisition) | — | — | — |
| Pre-2023 | Casetext venture stack (cumulative) | ~$70M | — | Union Square Ventures, Canvas, 8VC, Touchdown |
Thomson Reuters acquired Casetext on 17 August 2023 in an all-cash transaction valued at approximately $650M, per the TR press release. Pre-acquisition Casetext had raised approximately $70M across a venture stack led by Union Square Ventures, Canvas Ventures, 8VC and Touchdown Ventures (named-source coverage at LawSites / Artificial Lawyer). Post-acquisition the CoCounsel programme has been internally funded through Thomson Reuters’ Legal Professionals segment investment, which sits inside TR’s broader ~$11B capital capacity through 2028 and the $200M+/yr productized-AI commitment disclosed on Q4 2025 results. CoCounsel-specific spend is not separately disclosed.
Competitive Landscape
CoCounsel’s competitive set sits in three concentric rings: the symmetric legal-publishing incumbent (LexisNexis with Lexis+ AI / Protégé), independent legal-AI vertical plays (Harvey at the AmLaw 100 premium end, vLex on cross-border, Spellbook on contract drafting), and foundation-model providers offering general-purpose reasoning (OpenAI Deep Research is the most-cited unofficial substitute). CoCounsel and Harvey are the two most-named players when legal-AI analysts (LawSites, ALM Law.com, ABA Journal, Artificial Lawyer) discuss category leaders; CoCounsel and Lexis+ AI are the two corpus-owning incumbent activations of the category.
| Competitor | Positioning | Distribution edge | Threat profile |
|---|---|---|---|
| Harvey | Independent legal-AI vertical play backed by Sequoia, GIC, Kleiner Perkins and OpenAI Startup Fund. Embedded legal-engineering delivery into AmLaw 100 customer firms; ~$190M ARR and $11B valuation as of Mar 2026. | Direct to law-firm IT / GC procurement; Microsoft Word and Azure native; channel partnerships with DocuSign, LexisNexis and DeepJudge. | High — the main rival on capability and brand inside elite law-firm AI procurement; the head-to-head that defines the category. |
| Lexis+ AI / Protégé (LexisNexis (RELX, LON: REL)) |
Citation-verified research grounded in LexisNexis content with real-time Shepard’s validation. Protégé is the agentic layer launched in 2025 alongside Lexis+ AI. | Lives inside existing Lexis subscriptions across AmLaw 100, corporate legal and government — the symmetric distribution play against TR / CoCounsel. | High — the closest structural mirror in the cohort: corpus moat plus incumbent distribution, in the only other legal-publishing duopolist. |
| vLex / Vincent AI | International legal-AI play with a multi-jurisdiction case-law corpus (global coverage outside the US/UK Westlaw / Lexis duopoly); merged with Fastcase 2023. | Direct to firms and bar associations; cross-border emphasis; Vincent AI agentic layer launched 2024-2025. | Medium — differentiated on international coverage rather than head-to-head AmLaw 100 procurement; flanking risk on cross-border matters. |
| Spellbook | Word-integrated contract drafting and review for commercial lawyers; in-house and law-firm teams. 4,000+ team installations. | Microsoft Word native; self-serve and SMB-leaning go-to-market. | Medium — narrower (contracts) but stickier in that lane; competes on the contract-review use-case CoCounsel Drafting addresses. |
| OpenAI Deep Research / ChatGPT Enterprise (OpenAI) |
General-purpose research and reasoning agent — not legal-specialised but credible at legal research, and used unofficially by many attorneys. | Available to anyone with a ChatGPT subscription; no IT procurement required. | Medium-high and asymmetric — foundation-model provider competing at the workflow layer; the same indirect-competitor risk Harvey faces. |
Pricing benchmark: CoCounsel Core starts at $225 / user / month per published TR pricing; Harvey is reported at $1,000+ / user / month with one widely-cited figure of ~$40,000 / year for a 10-seat deployment; Lexis+ AI sits in the same band as CoCounsel; Spellbook is materially cheaper. CoCounsel competes on bundled Westlaw / Practical Law / KeyCite content access plus agentic workflow rather than on per-seat headline price.
Potential Risks
The case for CoCounsel at IM Framework 7.96 rests on the Westlaw + Practical Law + KeyCite corpus moat, TR’s installed-base distribution across AmLaw and corporate legal, the 1M-user milestone reached Feb 2026, and the segment’s contribution to the parent Legal Professionals 9% organic growth in Q4 2025. The case against splits into five risks of differing magnitude — with foundation-model supplier concentration and the Harvey head-to-head the most active, and the segment-of-conglomerate scoring discipline the most structural framing constraint.
Foundation-model supplier concentration — Claude Agent SDK is the agentic platform
CoCounsel’s foundation-model layer originally used OpenAI GPT-4 in the Casetext era and is now rebuilt on Anthropic’s Claude Agent SDK as announced November 2025. Single-FM-provider for the agentic platform is a real concentration risk, partly offset by TR’s institutional history of rapid FM-provider migration. The sub-rubric score on D4a supplier diversity is held at 6 on this evidence: multi-cloud at infrastructure level, but single-frontier-model for the agentic workflow surface. A capability slip at Anthropic, or a strategic shift in the partnership economics, would propagate directly to CoCounsel feature cadence.
Portability and substitutability at the matter-output level
CoCounsel matter outputs (memos, contract analyses, deposition prep) can be exported as documents, and prompts are portable in principle. But the customer’s value-stack — KeyCite citator graph, Practical Law templates, saved CoCounsel workflows, ONESOURCE+ tax integration — is captive to the Westlaw layer. Re-building equivalent workflows on Harvey or Lexis+ Protégé requires the firm to source an alternative authoritative corpus and rewire its knowledge management. The sub-rubric score on D1c portability is 5 on this evidence: standard exports work, but the analytical scaffolding does not move with the firm. The portability_weak Context Flag applies.
Harvey head-to-head on capability cadence
Harvey is the main independent rival and the most-named alternative in elite law-firm AI procurement. Harvey’s $11B valuation on ~$190M ARR (Mar 2026), the embedded legal-engineering delivery model, the multi-foundation-model routing (OpenAI, Anthropic, Google), and direct AmLaw 100 channel access are the principal capability threats. CoCounsel’s counter-position is the data moat (Westlaw / Practical Law / KeyCite) and the installed-base distribution; if Harvey closes the corpus access gap (via the LexisNexis case-law licence or an alternative), the structural advantage compresses materially.
Foundation-model provider encroachment — OpenAI Deep Research and ChatGPT Enterprise as indirect substitutes
Foundation-model providers continue to ship general-purpose research and reasoning agents at the ChatGPT Enterprise / Deep Research price point. Many attorneys reportedly use these unofficially. Bar association guidance is evolving but trending favourable to assisted research; the long-run question is whether unofficial use becomes procured use, and whether the corpus advantage is durable at the level of citation-grounded primary law. CoCounsel’s position depends on the corpus moat plus the agentic workflow integration; the harder threat is foundation-model providers building legal-specialised reasoning and citation-grounding directly into the base product.
Segment-of-conglomerate scoring — in-segment platform dynamics only
Per the IM Framework v1.5-A8 segment rule applied at scoring, D5 platform dynamics scores CoCounsel’s own platform dynamics — not Thomson Reuters’ broader Westlaw / Practical Law network. Within CoCounsel: emerging user-base network effects (1M users, workflow-template sharing in the 2026 reimagined release), but no multi-sided marketplace and a nascent complementor ecosystem (third-party integrations to CoCounsel API are early). The framework caps the segment’s D5 score at 6 on this evidence; the broader TR network effects flow through D3 distribution rather than D5.
Recent IM Coverage
Show recent press coverage of CoCounsel
- May 2026 — Thomson Reuters and Anthropic expand partnership to connect Claude with CoCounsel Legal via MCP integration.
- Feb 2026 — One million professionals turn to CoCounsel as Thomson Reuters scales agentic AI — 107 countries, three years after launch.
- Feb 2026 — Three years after launching as first AI legal assistant, CoCounsel reaches 1 million users.
- Feb 2026 — Thomson Reuters reports fourth-quarter and full-year 2025 results — Legal Professionals 9% organic growth, GenAI ACV share 28%.
- Nov 2025 — Thomson Reuters reveals new agentic AI in CoCounsel Legal — rebuilt on Anthropic’s Claude Agent SDK.
- Nov 2025 — Thomson Reuters announces upcoming agentic AI features for CoCounsel Legal.
- Jul 2025 — AmLaw 100 firm Polsinelli rolls out Thomson Reuters CoCounsel firmwide to 1,200 attorneys.
- Aug 2023 — Thomson Reuters completes acquisition of Casetext — $650M cash.
Curated feed of named-source coverage from approved publications — Thomson Reuters’ own news pages and Investor Relations releases, named-author legal-AI press (LawSites, Artificial Lawyer, Legal IT Insider, Law.com), and the segment’s earnings disclosures. We exclude PR-wire reposts of the same release and unsourced “industry round-up” pieces.
Show the source register for the figures on this page
IM operates a primary-source-where-possible discipline. The figures above come from:
- Revenue — segment basis: CoCounsel-specific revenue is not separately disclosed. The parent Thomson Reuters Legal Professionals segment delivered 9% organic growth in Q4 2025 with CoCounsel cited alongside Westlaw and Practical Law as a top driver per the Q4 2025 results announcement. GenAI-enabled share of TR annualised contract value reached 28% in Q4 2025, up from 15% when first disclosed in Q3 2024. Segment adjusted EBITDA margin was 47.3% for FY2025.
- Usage — 1 million professional users: Thomson Reuters disclosed 1M professional users across 107 countries in February 2026, three years after the March 2023 CoCounsel launch — see TR Investor Relations release and the LawSites report. Polsinelli firmwide rollout to ~1,200 attorneys with 70% associate / 50% shareholder active usage is the named AmLaw 100 deployment per Legal IT Insider.
- Headcount (basis-disclosure note): Thomson Reuters reports parent-level headcount in its annual report (~26,000 employees globally as of the most recent disclosure); CoCounsel and the absorbed Casetext team are not broken out separately. We decline-to-publish a precise CoCounsel-segment headcount figure pending a primary disclosure, and reference only that the Casetext engineering team was absorbed in 2023 and that TR has expanded the legal-AI engineering investment alongside the Nov 2025 Claude Agent SDK rebuild.
- Funding to date: Thomson Reuters acquired Casetext for $650M cash on 17 August 2023 per the TR press release. Post-acquisition, the CoCounsel programme is funded internally inside TR’s Legal Professionals segment investment, which sits within TR’s broader ~$11B capital capacity through 2028 and the $200M+/yr productized-AI commitment disclosed on Q4 2025 results. CoCounsel-specific spend is not separately disclosed.
Methodology & Disclaimer
For metric definitions, source-tier hierarchy, and decline-to-publish rules, see the tracker methodology. Confidence dots (• green / • amber / • red) follow the same convention as the AI Tracker.
Spotted a figure you believe is wrong? Send corrections to info@informationmatters.net.
Information Matters Framework scores are the considered opinion of the IM team — human and AI — applied to publicly-available evidence under a disclosed methodology. They are not statements of fact about the companies scored and they are not investment advice.
