AI legal assistant built on GPT-4 for legal research and document review
Casetext is an AI-powered legal research and document analysis platform for lawyers and legal professionals.
AI Panel Score
6 AI reviews
Reviewed
AI Editor ApprovedApproved and published by our AI Editor-in-Chief after full panel analysis.Casetext's core product, CoCounsel, functions as an AI assistant that legal professionals interact with by assigning specific tasks rather than typing open-ended prompts. Users select a skill, such as document review, legal research memo, deposition preparation, or contract review, upload relevant files or describe a legal question, and CoCounsel returns organized output with citations back to source documents or case law. This task-based structure is designed to fit into existing workflows like discovery review or memo drafting rather than requiring lawyers to learn prompt engineering.
The platform highlights several named skills as distinguishing features: Document Review for flagging responsive or privileged documents across large sets, Legal Research Memo for generating research summaries with citations, Deposition Preparation for building outlines from case documents, Contract Review for identifying deviations from playbooks or missing clauses, and Contract Policy Compliance for checking documents against internal policies. Casetext also integrates with legal databases for case law and statutes, and it emphasizes that every output includes citations that can be checked against source material to reduce reliance on unverified AI-generated text.
Casetext is aimed at law firms, corporate legal departments, and individual attorneys who handle research-intensive or document-heavy matters such as litigation, contract review, and compliance work. Pricing is not publicly listed on a self-serve basis; prospective customers contact Casetext for quotes, which typically depend on firm size and usage volume. It competes with other AI legal research and drafting tools such as Harvey, Lexis+ AI, and Westlaw's AI-assisted research products.
Casetext operates as a web-based platform accessed through a browser, with document upload and review handled within its interface. It does not publish a public API for third-party integration, and it is positioned primarily as an end-user application for legal teams rather than a developer platform.
An AI-powered brief analysis tool, recognized with the World Economic Forum's Technology Pioneer award, that helped establish Casetext's legal AI innovation prior to CoCounsel.
Identifies relevant topics to address in a deposition and drafts targeted questions for each, producing a refined deposition outline based on user-specified areas of interest.
Lets users converse with CoCounsel in plain English to delegate substantive legal tasks, powered by OpenAI's GPT-4 large language model that has passed the Uniform Bar Exam.
Generates comprehensive legal research memos with citations in response to natural-language legal questions, such as researching whether protected employee activities lose protection under Title VII.
Analyzes a specific contractual provision against typical market norms to help attorneys determine whether a clause is standard for its type of agreement.
Reviews uploaded contracts against a stated company policy, flags noncompliant clauses (e.g., arbitration exemptions), and recommends redlines downloadable in Microsoft Word.
Drafts correspondence tailored to different audiences, including clients, judges, opposing counsel, and third parties.
Searches an extensive, up-to-date collection of caselaw and statutes and cites sources in its outputs so results can be efficiently verified by attorneys.
Rapidly analyzes large sets of documents for use cases like due diligence, e-discovery, and internal investigations, quickly identifying key documents and extracting critical information.
Provides a menu of purpose-built application "skills" (e.g., legal research, document review, contract analysis) that users can launch directly from the chat interface, with recently used skills surfaced automatically.
Integrates CoCounsel's AI capabilities across Thomson Reuters products including Westlaw Precision, Practical Law, Document Intelligence, and HighQ, while also remaining available as a standalone product.
Encrypts all content processed through the platform end-to-end and ensures client materials and data are never stored or used to train the underlying AI models.
Casetext (now part of Thomson Reuters as CoCounsel) does not publish self-serve pricing. It is sold to law firms and legal departments on a custom, sales-led basis depending on firm size, seat count, and module selection (legal research, AI drafting/review, etc.); prospective buyers must contact Thomson Reuters/CoCounsel sales for a quote.
Thomson Reuters paid $650 million for this. That's the whole pitch.
“Casetext got acquired before it could fail as a startup. Now it's a feature line inside a company that isn't going anywhere.”
Vendor risk question answered on day one: Thomson Reuters bought Casetext in June 2023 for $650 million. That's not a startup you're betting on anymore, that's an incumbent's product roadmap. Founded 2013, a decade in market before acquisition — this isn't a two-year-old app hoping for a Series C.
The skills menu is the smart part: Deposition Preparation, Contract Policy Compliance, Legal Research Memo. Task-based, not prompt-based, which matters for a profession that bills by the hour and doesn't want to learn prompt engineering. Citations back to source case law address the hallucination problem that scares every general counsel.
No public pricing, sales-led quotes only, and it now competes with Harvey and Lexis+ AI, both well-funded and moving fast. Adopting this reads as safe to a board, not bold. Fine for research-heavy litigation shops, less exciting if you want to look ahead of the market.
Competes directly with Harvey and Lexis+ AI, both moving aggressively in the same space.
Backed by a known legal publisher, low risk to defend to a board.
Task-based skills like Document Review fit existing discovery workflows with minimal ramp.
Speeds up existing research and review workflows rather than creating new capability.
Owned by Thomson Reuters since 2023 for $650 million; not going anywhere.
Litigation and corporate legal teams that want AI research tools backed by an established legal publisher.
Skip if you need API access or self-serve pricing to move fast.
Sound outside counsel tool, but now a Thomson Reuters product with all the platform lock-in that implies.
“CoCounsel's skill-based structure fits how litigators and deal lawyers actually delegate work. The $650M Thomson Reuters acquisition changes the governance calculus more than the technology does.”
The task-based design is the right call for privilege and work-product concerns. Document Review, Deposition Preparation, Contract Policy Compliance — these map onto discovery and diligence workflows attorneys already run, not open-ended chat that invites hallucination risk. Citations tied back to source documents give me something I can actually defend to a client or a judge under Rule 11.
The 2023 Thomson Reuters acquisition is the fact I'd flag to the partnership. Now bundled into Westlaw Precision, Practical Law, and HighQ, meaning our research stack and our AI assistant share one vendor. If that vendor changes pricing or terms, we have no Lexis+ AI or Harvey fallback without a re-procurement cycle.
No public API, no self-serve pricing — sales-led, quote-based, opaque for budgeting. Encryption and no-training-on-client-data claims are the right representations for outside counsel guidelines, but I'd want that in the MSA, not the marketing copy.
Sits alongside Harvey and Lexis+ AI as a category leader with $650M acquisition validation behind it.
Task-based skill selection mirrors how associates delegate discrete assignments rather than free-text prompting.
Cross-product tie-in with Westlaw Precision and Practical Law is convenient but no public API limits third-party DMS or workflow integration.
Post-acquisition, adoption ties research and AI review to one Thomson Reuters vendor stack, a concentration risk over a 3-year contract term.
Five named skills (Document Review, Legal Research Memo, Deposition Prep, Contract Review, Contract Policy Compliance) show real workflow modeling, not a chatbot wrapper.
Firms and legal departments already committed to the Thomson Reuters/Westlaw ecosystem needing document review and research support.
Avoid if your firm requires multi-vendor redundancy or a published API for internal system integration.
No sticker price. $650M acquisition price tag, zero pricing transparency for buyers.
“Casetext, now CoCounsel under Thomson Reuters, publishes zero self-serve pricing. Every dollar figure requires a sales call, which means every TCO model starts as a guess.”
Contact Sales. That's the only tier listed. No per-seat number, no volume breakpoints, nothing procurement can model before a call.
Thomson Reuters bought Casetext in 2023 for $650M. That price gets baked into quotes somehow — bundled seat, Westlaw integration, module selection all move the number. A 50-attorney firm has no way to estimate year-1 spend, let alone year-3 TCO with Document Review and Contract Policy Compliance add-ons stacked on. Compare that to Lexis+ AI or Harvey, both also sales-led — category norm here is opacity, not just Casetext.
ROI has a real anchor: hours saved on document review and deposition prep are billable-hour equivalents, easy to track against invoices. But without base pricing, you can't compute payback period. No published free trial either, so procurement eats the negotiation cost just to get a number to model against.
Sales-led onboarding under Thomson Reuters likely means legal, security review, and IT vetting before seat one is live.
No public terms on renewal or cancellation; enterprise legal contracts typically run 12+ months with negotiation room only at scale.
Single 'Contact Sales' tier, no published rates, category norm but still a black box.
Time-saved on document review and memos maps cleanly to billable hours, a measurable proxy even without pricing.
Custom pricing by firm size and module means 3-year cost can't be modeled without a quote.
Firms with procurement teams willing to negotiate custom Thomson Reuters bundles.
Avoid if you need a comparable number before your first sales call.
Skills menu, not a chatbox — the way I actually want to hand off work.
“CoCounsel's task-based skills fit how a paralegal actually gets assigned work: document review, deposition outlines, memo drafts. The citation-back-to-source habit is the feature that matters most for my day-to-day sanity check.”
The skills menu is the right call. Document Review, Deposition Preparation, Contract Policy Compliance — these map to actual assignments partners hand down, not open-ended prompt writing I'd have to learn on my own time. Every output cites back to source documents or case law, which means I'm verifying, not trusting blind. That's the daily habit that makes this usable instead of scary.
Friction shows up in the parts I can't see before buying. No self-serve pricing, no free trial — I'm stuck waiting on a sales quote depending on firm size and seat count, same as Harvey or Lexis+ AI. Deposition Preparation drafting outlines from case documents saves real hours, but I still need to read every flagged clause in Contract Review myself.
No public API, no changelog, no docs page listed in the evidence. For a $650 million Thomson Reuters acquisition, that's thin visibility into how the tool actually improves week to week. Fine for firms with dedicated support; frustrating if you're the one troubleshooting alone.
Citation-backed outputs mean verification is built in, not an afterthought after the novelty wears off.
Evidence shows no docs page, changelog, or API reference — hard to judge how skills are taught to new users.
No free trial and no published pricing means friction starts before you even open the tool.
Market Check and Contract Policy Compliance suggest depth beyond basic search, though advanced skill discoverability isn't detailed.
Task-based skills like Document Review and Deposition Preparation mirror how assignments actually arrive, per the description.
Litigation and contract-heavy legal teams that already work inside Thomson Reuters products like Westlaw or Practical Law.
You need transparent self-serve pricing or a free trial before committing firm budget.
A skills menu instead of a blank chat box, which is the smartest thing about it.
“Casetext skips the blinking cursor problem by making you pick a task first. That structure matters more than people give it credit for.”
No pricing page you can actually click through, no free trial, nothing self-serve. You call sales, they figure out your seat count, you wait for a quote. For a firm that's fine, that's how legal software always worked. For a solo attorney kicking tires, that's homework before you've even seen the product.
What I like is the Skills Menu. Document Review, Deposition Preparation, Contract Policy Compliance — named tasks, not a naked prompt box. That's the difference between asking a lawyer to learn prompt engineering and asking them to click a button they already understand. Compare that to a raw GPT-4 chat window and it's obvious which one a partner actually uses at 6pm on a Friday.
Mobile isn't mentioned anywhere, and legal review work happening only at a desk isn't shocking, but it does mean this is a desk tool, full stop. Citations back to source documents are the real trust feature here — that's what stops this from being another Harvey-style black box. Thomson Reuters bought this for $650 million in 2023, so the money says the workflow works. Whether it feels like a product you enjoy using three months in is a different question nobody's answered publicly yet.
Named skills like Market Check and Correspondence Drafting suggest real workflow thought, though no public docs or changelog to confirm ongoing sweat.
Task-based skills menu instead of open prompting is designed specifically to flatten the learning curve for non-technical lawyers.
Platforms listed as web only, no mobile mention anywhere in the evidence.
No self-serve trial or pricing means the first ten minutes happen in a sales call, not the product itself.
Citation-backed outputs and stated no-training-on-client-data policy signal a team building for trust, not demo glow.
Law firms and corporate legal departments with document-heavy litigation or compliance work who can commit to a sales-led contract.
You're a solo attorney or small shop wanting to test-drive pricing and features before talking to a sales rep.
$650 million acquisition is the strongest data point here, not the demo.
“Thomson Reuters bought Casetext in June 2023 for $650 million — that's the viability signal that matters. Product depth looks real, but you're now betting on Thomson Reuters' roadmap, not a startup's.”
No self-serve pricing. Contact sales, quote depends on seat count. Category norm for enterprise legal tools, but it means no one outside a firm's procurement office actually knows what this costs.
The acquisition changes the read. Founded 2013, CARA AI won a World Economic Forum award years before CoCounsel existed — this isn't a ChatGPT wrapper thrown up in 2023. Thomson Reuters paid $650 million for it and folded it into Westlaw Precision, Practical Law, HighQ. That's real integration, not a badge partnership.
Tradeoff: no public API, no docs, no changelog visible. You're inside the Thomson Reuters ecosystem or you're not. Competitors here are Harvey and Lexis+ AI — both well-funded, both chasing the same law firms. Casetext's differentiation is the case law citation grounding, which addresses the hallucination problem lawyers actually fear. Exit story is the real question: once embedded in Westlaw workflows, unwinding gets expensive.
Crowded field with Harvey and Lexis+ AI, but citation-grounding to source docs is a real, stated wedge.
No public API, web-only, deep Westlaw/Practical Law integration makes unwinding costly.
Thomson Reuters acquisition for $650M in 2023 is about as strong a viability signal as this category gets.
Citation-back-to-source claim is specific and checkable, not vague AI superlative.
Pre-dates the GPT-4 hype cycle — CARA AI and WEF recognition predate CoCounsel by years.
Mid-size to large firms already using Westlaw or Practical Law who want AI folded into existing workflow.
You're a solo practitioner or small firm wanting transparent pricing and quick self-serve setup.
Common questions answered by our AI research team
CoCounsel performs document review, legal research memos, deposition preparation, and contract analysis.
Yes, CoCounsel is trained specifically on legal work product and case law databases.
Yes, deposition preparation is one of CoCounsel's core tasks.
Yes, contract analysis is one of CoCounsel's core tasks.
CoCounsel is built on large language models trained specifically on legal work product and case law databases.
Casetext is a San Francisco-based legal technology company that develops AI-powered legal research, drafting, and document analysis tools for law firms and corporate legal departments.
