Casetext AI Is Gone: How Its Technology Now Powers Thomson Reuters' CoCounsel

Casetext AI, once one of the most closely watched startups in legal technology, no longer exists as an independent product. Thomson Reuters acquired the company in 2023 for $650 million and shut down the standalone Casetext platform on April 1, 2025. For anyone still searching for "Casetext AI" expecting to sign up for the service, the short answer is that it has been folded entirely into Thomson Reuters' CoCounsel, the company's flagship AI legal assistant.

The shutdown was not framed as a failure. Casetext's underlying technology, built on GPT-4 class models and grounded legal citations, was the first legal AI tool to demonstrate GPT-4 level performance on real legal tasks, and Thomson Reuters has continued investing in it rather than discontinuing the capability. The brand disappeared; the engineering did not.

What Happened to Casetext AI

Casetext built its reputation on CoCounsel, an AI legal assistant that could conduct legal research, review and summarize documents, prepare deposition outlines, identify relevant contract clauses, and draft legal memos, tasks that previously consumed hours of associate attorney time. Thomson Reuters, the parent of the legal research platform Westlaw, bought Casetext in 2023 specifically to bring that assistant in-house.

Rather than run Casetext as a separate product line indefinitely, Thomson Reuters spent roughly two years integrating its research and reasoning capabilities into its own CoCounsel platform and into Westlaw itself. The standalone Casetext website and login were retired on April 1, 2025, redirecting existing customers to CoCounsel.

Key Details

CoCounsel today is positioned as a Thomson Reuters product, available either as a standalone subscription or bundled with a Westlaw subscription. Pricing depends on firm size and usage volume, and is negotiated directly with Thomson Reuters' sales team rather than published as a flat rate, which is standard practice for enterprise legal software aimed at law firms and corporate legal departments.

Functionally, CoCounsel retains the core Casetext capabilities: grounded case-law research with citations back to primary sources, contract analysis, deposition preparation, and document review at a scale that would be impractical for a human associate working alone. The emphasis on grounded citations, rather than open-ended generative answers, remains central to how Thomson Reuters markets the tool to a profession where an unverifiable AI citation can mean a sanctioned attorney.

Why It Matters

The Casetext-to-CoCounsel transition is a useful case study in how AI acquisitions in specialized, high-stakes industries tend to play out. Legal research demands accuracy and auditability that general-purpose chatbots do not need to provide, and Thomson Reuters' strategy was to absorb the acquired team and technology into an existing, trusted enterprise brand rather than let two competing legal AI products confuse the market.

For law firms and legal departments that adopted Casetext specifically for its independence from the major legal database incumbents, the shutdown also illustrates a common risk in relying on AI startups: acquisition by an established player can mean better resourcing and integration, but it also means the product a customer chose no longer exists as such.

Industry Context: Legal AI's Consolidation Phase

Casetext's absorption into Thomson Reuters mirrors a broader pattern across legal AI. Competing tools such as Harvey and NexLaw continue to market themselves as CoCounsel alternatives, explicitly positioning against Thomson Reuters' now-dominant offering. Legal AI vendor comparisons published throughout 2026, alongside a broader wave of emergent enterprise AI capability across other industries, routinely benchmark Westlaw, LexisNexis and CoCounsel-class large language model tools against one another, reflecting a market that has consolidated around a handful of well-capitalized incumbents rather than the fragmented startup landscape of 2022 and 2023.

That consolidation raises real questions for smaller legal AI vendors: differentiate sharply on price, niche practice areas or workflow integration, or risk being acquired and absorbed the way Casetext was.

What It Means for Legal Teams Evaluating AI Tools

For firms currently evaluating legal AI, the practical lesson is to search for the product's current owner and roadmap before adopting it, not just its feature set. A tool built by a well-funded independent startup can become a legacy brand name inside a much larger company within two years, changing pricing, support and integration paths along the way.

It is also worth noting that the underlying research quality that made Casetext notable, grounded citations checked against real case law, did not disappear. If anything, CoCounsel's integration with Westlaw's existing case-law database likely improved the grounding available to the same underlying technology.

Why Grounded Citations Matter So Much in Legal AI

The emphasis on grounded, verifiable citations is not a minor technical detail; it is arguably the single most important design requirement in legal AI. Multiple attorneys across different firms and jurisdictions have faced sanctions or public embarrassment after submitting court filings containing AI-generated case citations that turned out to be entirely fabricated, cases that do not exist, quotes that were never written, or holdings that misstate what a real court actually decided. Casetext's original pitch, and CoCounsel's pitch today, is built specifically around avoiding that failure mode by tying every citation back to a real, retrievable primary source rather than letting a language model generate references from memory.

That grounding requirement is also why legal AI has developed differently from consumer chatbots. A general-purpose assistant can be useful even when it occasionally produces a plausible-sounding but incorrect answer, since the cost of an error is usually low. In legal research, an ungrounded hallucination can mean a wasted brief, a missed deadline, or a sanctioned attorney, which is why Thomson Reuters continued investing in the underlying Casetext technology rather than treating it as a commodity feature to bolt onto Westlaw with less rigor.

The Broader Legal AI Market in 2026

Casetext's absorption into CoCounsel has not ended competition in legal AI; if anything, it has sharpened it. Competing platforms including Harvey and NexLaw actively market themselves as CoCounsel alternatives, arguing they offer different strengths in areas such as litigation-specific workflows or pricing flexibility for smaller firms. Industry comparisons published through 2026 routinely benchmark Westlaw, LexisNexis and CoCounsel-class large language model tools side by side, reflecting a market where a handful of well-capitalized platforms, each backed by an established legal publisher or a well-funded startup, now compete primarily on integration depth, citation reliability and firm-specific workflow support rather than raw model capability.

For law firms making a purchasing decision today, that competitive landscape means evaluating not just a tool's current feature set, but its financial backing, release cadence and long-term product roadmap, since the Casetext story shows how quickly even a well-regarded standalone product can be absorbed into a larger platform.

Final Takeaway

Casetext AI is gone as a brand, not as a capability. Anyone encountering the name today should know it now refers to CoCounsel, available through Thomson Reuters and Westlaw, and that the underlying legal research technology Casetext pioneered continues to be actively developed rather than shelved.

FAQs