CoCounsel started life as Casetext, one of the first legal tech companies to build a real product on GPT-4. Thomson Reuters acquired Casetext in 2023 and folded CoCounsel into its stack, and today it is Thomson Reuters’ flagship legal AI assistant. The thing that separates it from a general chatbot is where its answers come from: CoCounsel is grounded in Westlaw and Practical Law content rather than the open web. That grounding is the whole pitch, and it is a strong one for firms already living inside the Thomson Reuters ecosystem.
CoCounsel is Thomson Reuters’ legal AI assistant, originally built by Casetext, that runs skills like legal research, document review, and contract analysis. Its edge is grounding answers in Westlaw and Practical Law rather than the open web, which reduces hallucination risk. Pricing is sales-led and quote-based, now largely bundled into Westlaw tiers. Best for Westlaw firms wanting a research-grounded assistant.
Faz says: The pitch that actually matters here is grounding. A raw chatbot invents case citations because it is predicting text, not checking a database. CoCounsel ties its skills to Westlaw and Practical Law, so its research runs against real, editorially maintained legal content. That reduces hallucination risk, it does not eliminate it, and you still have to verify every citation before it goes in a brief. But for a firm that already pays for Westlaw, having the AI reason over that same authoritative content instead of the open internet is a genuine advantage. Just do not treat it as a lawyer. Treat it as a very fast junior associate whose work you always check.
Saru says: This review draws on Thomson Reuters’ official CoCounsel documentation, its published product tiers, the Casetext acquisition record, and reported pricing, current to 2026. CoCounsel pricing is sales-led and not fully public, so the numbers here are historical and directional. Get a written quote for your firm size and Westlaw configuration before you buy, because bundling and per-seat pricing change.
This post may contain affiliate links. If you buy through them we earn a small commission at no extra cost, and it never changes our view.
Quick Facts
| Tool | CoCounsel (Thomson Reuters) |
|---|---|
| Category | Legal AI assistant |
| Best for | Westlaw and Practical Law firms wanting a research-grounded assistant |
| Core “skills” | Legal research, review documents, summarize, extract contract data, prepare for a deposition, timeline |
| Grounding source | Westlaw and Practical Law content |
| Pricing | Sales-led, quote-based; historically around $225/user/mo standalone, now bundled with Westlaw tiers |
| Our score | 4.2 / 5 |
What CoCounsel Actually Does
CoCounsel is organized around “skills,” which are named, purpose-built tasks rather than a single open-ended chat box. The core set covers the work that eats the most associate hours:
- Legal research: ask a question in plain English and get an answer with citations, run against Westlaw content rather than the open web.
- Review documents: upload a set of documents and ask questions across all of them at once, which is the classic discovery and due-diligence grind.
- Summarize: condense long documents, contracts, or transcripts into readable summaries.
- Extract contract data: pull key terms, clauses, and data points out of contracts at volume.
- Prepare for a deposition: generate deposition questions and outlines from your materials.
- Timeline: build a chronology of events from a document set, which is tedious to do by hand.
The differentiator across all of these is grounding. CoCounsel’s research and drafting are tied to Westlaw and Practical Law, the same editorially maintained content Thomson Reuters has built for decades. The original CoCounsel was built by Casetext on OpenAI’s GPT-4, and Thomson Reuters has since evolved the platform, with the next generation of CoCounsel Legal built on Anthropic’s Claude models. The underlying model matters less than the fact that its answers are checked against a real legal database, which is what separates this from asking a general assistant to “find me a case.”
The Casetext Backstory
CoCounsel did not start at Thomson Reuters. Casetext, founded in 2013, built the original CoCounsel on GPT-4 after getting early access to the model from OpenAI. It was one of the first credible legal AI assistants on the market, used by thousands of law firms and legal departments.
Thomson Reuters acquired Casetext in 2023 for a reported $650 million in cash, one of the largest deals in legal tech to that point. Since then, CoCounsel has been woven through the Thomson Reuters stack. You now see it as a standalone assistant (branded over time as CoCounsel Core and CoCounsel Legal) and embedded directly inside Westlaw as Westlaw Precision with CoCounsel. The practical upshot: “CoCounsel” today is less a single product and more a layer of AI skills that shows up across Thomson Reuters’ legal tools, all sharing the same grounded-in-Westlaw approach.
Grounded Answers and Hallucination Risk
Here is why grounding is worth caring about. A general chatbot generates text by predicting what comes next, which is how you end up with confident, well-formatted, completely fake case citations. That is not a hypothetical: lawyers have been sanctioned for filing briefs with invented cases produced by general AI tools.
Tying an AI’s answers to a legal database changes the failure mode. When CoCounsel runs legal research, it is retrieving and reasoning over real Westlaw content rather than pulling citations out of a language model’s memory. That makes fabricated citations far less likely and makes the answer traceable back to a source you can open and read.
What it does not do is make the tool infallible. No legal AI is hallucination-free. A grounded system can still misread a holding, over-generalize, miss a controlling authority, or summarize a case in a way that shades the meaning. The correct posture is unchanged: CoCounsel speeds up the first draft of research and review, and you verify every output before it leaves your desk. It reduces risk. It does not transfer responsibility off the attorney.
Pricing
This is the honest part. CoCounsel pricing is sales-led and not fully public, so treat anything you read, here included, as directional.
Historically, the standalone Casetext CoCounsel assistant was priced around $225 per user per month. As Thomson Reuters has integrated it, pricing has shifted toward bundling into Westlaw subscriptions rather than a clean standalone number. Reported configurations put Westlaw Precision with CoCounsel in the neighborhood of $400-plus per user per month once you combine the AI assistant with the Westlaw research it needs to be useful, and multi-year contracts carry discounts. Those figures move by firm size, jurisdiction coverage, and contract term.
The practical takeaway: do not budget off a blog post, this one included. Get a written quote for your exact seat count and Westlaw configuration. Ask specifically whether case law search is included or whether it requires a separate Westlaw Precision subscription on top, because that single question changes the real total.
Where It Fits
CoCounsel makes the most sense for firms already inside the Westlaw and Practical Law ecosystem. If you pay for Westlaw and your team lives in it, layering CoCounsel’s skills onto that content is a natural fit, and the grounding advantage is real because it is your existing research corpus.
It fits best at midsize firms and up, where the volume of document review, contract analysis, and research justifies the spend and the sales-led procurement process. Large firms and legal departments with heavy discovery and due-diligence loads get the most leverage.
Who should look elsewhere: solos and small firms on tight budgets, and anyone not committed to Westlaw. If price is the deciding factor, cheaper research-grounded assistants like Paxton or vLex’s Vincent AI cover a lot of the same ground for less, and they do not assume you are already paying for the Thomson Reuters stack. The CoCounsel value proposition weakens sharply if you are buying Westlaw purely to justify the AI on top.
Pros and Cons
Pros
- Answers grounded in Westlaw and Practical Law, not the open web, which reduces hallucination risk
- Purpose-built skills for research, document review, contract extraction, and deposition prep
- Strong pedigree: Casetext was an early, credible GPT-4 legal product, now backed by Thomson Reuters
- Deep integration with Westlaw for firms already in that ecosystem
- Traceable citations you can open and verify
Cons
- Pricing is sales-led, quote-based, and not fully public
- Real cost is high once you bundle the Westlaw research it depends on
- Value proposition is weak if you are not already a Westlaw firm
- Still requires human verification of every output, like all legal AI
- Overkill for solos and small firms where cheaper tools cover the core need
How It Compares
Against Lexis+ AI, the choice mostly comes down to which research platform your firm already uses. Lexis+ AI grounds its answers in LexisNexis content the same way CoCounsel grounds in Westlaw, so a LexisNexis firm will usually prefer Lexis+ AI and a Westlaw firm will prefer CoCounsel. The grounding philosophy is similar; the ecosystem lock-in decides it.
Against Harvey, the difference is horizontal versus grounded. Harvey is a frontier-AI legal platform aimed at large firms, strong on drafting and general reasoning across a firm’s own documents, and less tied to a single research database. CoCounsel leans harder on grounding in Westlaw’s authoritative content. Our Harvey AI review goes deep on where that trade-off lands.
Against budget tools like Paxton or vLex’s Vincent AI, CoCounsel is the premium, ecosystem-integrated option. Paxton delivers a lot of the same grounded research and drafting for a fraction of the price and without assuming a Westlaw subscription, which makes it the better call for cost-sensitive firms. We line all of these up in our best AI legal research tools guide.
FAQ
What is CoCounsel?
CoCounsel is Thomson Reuters’ legal AI assistant. It runs skills like legal research, document review, summarization, contract data extraction, deposition prep, and timeline building, with its answers grounded in Westlaw and Practical Law content rather than the open web. It began as a Casetext product and is now integrated across the Thomson Reuters stack, including inside Westlaw.
Is CoCounsel the same as Casetext?
Effectively, yes, in origin. Casetext built the original CoCounsel on GPT-4, and Thomson Reuters acquired Casetext in 2023 for a reported $650 million. The CoCounsel brand lives on as Thomson Reuters’ legal AI assistant, now appearing as CoCounsel Core, CoCounsel Legal, and embedded in Westlaw Precision with CoCounsel. Casetext as a standalone brand has been absorbed into Thomson Reuters.
How much does CoCounsel cost?
Pricing is sales-led and not fully public, so treat any number as directional. Historically the standalone assistant ran around $225 per user per month. Thomson Reuters now largely bundles CoCounsel into Westlaw subscriptions, and reported combined configurations land north of $400 per user per month once you include the Westlaw research it depends on. Get a written quote for your firm.
Does CoCounsel hallucinate?
Less than a general chatbot, but not never. Grounding its answers in Westlaw content makes fabricated citations far less likely, because it retrieves real sources rather than inventing them. No legal AI is hallucination-free, though. CoCounsel can still misread a holding or miss controlling authority, so you must verify every output before relying on it.
Is CoCounsel better than Lexis+ AI?
Neither is universally better; it depends on your research platform. CoCounsel grounds in Westlaw and Practical Law, while Lexis+ AI grounds in LexisNexis content. A Westlaw firm will generally get more from CoCounsel, and a LexisNexis firm will generally prefer Lexis+ AI. The deciding factor is which ecosystem you already pay for.
Is CoCounsel worth it?
For midsize-and-up firms already inside the Westlaw and Practical Law ecosystem, yes: the grounding is real and the skills save meaningful associate time. For solos, small firms on tight budgets, or firms not committed to Westlaw, the sales-led pricing and ecosystem dependence make cheaper grounded tools like Paxton a better-value choice.
Verdict
CoCounsel is a genuinely strong legal AI assistant, and it earns a 4.2 out of 5. Its grounding in Westlaw and Practical Law is the real deal: it reduces hallucination risk in a way that matters for legal work, and its skills cover the highest-volume associate tasks well. The pedigree is solid, from an early GPT-4 product at Casetext to Thomson Reuters’ flagship AI layer.
What holds it back from a higher score is not the product, it is the packaging. Pricing is sales-led and opaque, the real cost is high once you fold in the Westlaw research it needs, and the value proposition collapses if you are not already a Westlaw firm. Buy it for the right firm, in the right ecosystem, and it is excellent. If you are shopping on price or platform-agnostic, compare the field in our best AI legal research tools guide and the wider stack in our best AI tools for lawyers pillar.
More AI tool guides worth reading: Best Legal Document Automation Software and Best AI Tools for Paralegals.
Written by
FazFaz is the founder of AIToolsBakery. Every tool on this site is personally tested with real-world writing tasks before a single word gets published. No sponsored rankings, no recycled press releases.
Read more about how we test →
