The best AI for lawyers in 2026

This guide is for partners and owners of small and mid-size firms who want to know which AI tools are worth paying for. Law is different from most industries in two ways: you owe clients confidentiality under Rule 1.6, and you sign what you file. A tool that invents a case citation can get you sanctioned, so the picks below lean toward products grounded in real legal databases and vendors that publish clear data-use terms.

Quick comparison

How we picked

• We reviewed vendor documentation, published pricing pages (checked November 2026), security and data-use terms, and user reviews on G2, Capterra and Reddit.
• We did not run hands-on tests. Where a vendor does not publish a price, we say so instead of repeating third-party estimates, which vary widely.
• We weighted tools that cite real legal sources, state that customer data is not used to train models, and fit firms of roughly 2 to 50 lawyers.
• We grouped tools by the job they do, because most firms need more than one.

Legal research and drafting

CoCounsel Legal (Thomson Reuters)

CoCounsel Legal takes a plain-language request, builds a plan, and runs research, analysis and drafting against Westlaw, Practical Law and your own documents, with linked citations. It also includes Westlaw Brief Builder and a tabular review feature that Thomson Reuters says handles up to 10,000 documents against 100 questions. Thomson Reuters states it is SOC 2 and ISO 42001 certified and that customer data is never used to train the underlying models.

• Best for: Litigation and research-heavy firms already on Westlaw.
• Pricing: Not published. “View plans” and sales contact only; a free trial is available on request.
• Watch out for: It is strongest when you also pay for Westlaw, so budget for both. Vendor ROI figures on its site come from a study Thomson Reuters commissioned.

Lexis+ with Protégé

LexisNexis renamed Lexis+ AI to Lexis+ with Protégé in early 2026. Protégé is the assistant inside the platform and is not sold on its own. Its most useful feature for risk control is that it checks citations in AI-generated and attorney-drafted text against LexisNexis sources and flags any it cannot verify, alongside Shepard’s treatment signals. You can load matter documents into Vaults for timelines, counterarguments and drafts.

• Best for: Firms on Lexis who want built-in citation checking.
• Pricing: Custom, based on firm size and content. LexisNexis says solo and 1-3 attorney firms can buy some Lexis+ packages online or request a quote. Free 2-day trial.
• Watch out for: The trial is short, so line up real research questions before you start it.

Spellbook

Spellbook works inside Microsoft Word to draft, redline and review contracts. The Law Firms package includes the Word add-in and Spellbook’s Associate agent, group training, and dedicated support for teams over 10.

• Best for: Transactional, real estate, business and estate planning practices that live in Word.
• Pricing: Not published; priced by number of seats after a demo. 7-day free trial.
• Watch out for: It is a drafting tool, not a research database. Verify any legal authority it suggests.

A note on Harvey

Harvey is the best-known legal AI platform, but it is built and priced for large firms. Its site cites 80+ AmLaw 100 customers, lists no pricing and offers demos only. It does have a mid-size firm page. If you are under about 25 lawyers, the tools above are a more realistic fit.

Practice management with built-in AI

For many small firms, the cheapest first step is turning on AI inside the system you already pay for. It sees your matters, contacts and documents without any extra plumbing.

Clio Manage and Clio Work

Clio Manage plans start at $49 per user per month for Starter, and Clio says AI is included from the Core plan up. Core, Signature and Elite are quote-only in the US. Clio Work is a separate AI workspace for research, analysis and strategy, powered by the vLex legal library Clio acquired in 2025. It launched at $199 per user per month and, since April 2026, can be bought without Clio Manage.

• Best for: Small firms already on Clio, or ones choosing a practice management system now.
• Pricing: Manage Starter $49/user/month; higher tiers by quote. Clio Work launched at $199/user/month; confirm the current standalone price with Clio. 7-day free trial.
• Watch out for: Clio has changed plan names and AI packaging more than once, so get the AI features you need written into your quote.

MyCase (8am IQ)

MyCase is one of the few vendors with full public pricing. Pro ($100/user/month billed annually) includes the 8am IQ writing and document assistant. Advanced ($130) adds the Case Assistant for case analysis and the Discovery Assistant, which OCRs 5,000 pages per user per month (extra pages are $30 per 5,000). MyCase also lists an MCP connector that lets you reach MyCase data from Claude.

• Best for: Small firms that want predictable costs.
• Pricing: Basic $50, Pro $100, Advanced $130 per user per month billed annually ($60/$120/$150 monthly). 10-day free trial, no card required.
• Watch out for: Basic has no AI features, and the strongest tools are on Advanced only.

Smokeball

Smokeball builds AI into client communication on its Boost, Grow and Prosper+ plans. AI-powered intake is included on Prosper+ and sold as an add-on on Boost and Grow. Its Archie AI assistant is an add-on on Boost and above.

• Best for: Small firms that want AI in client communication and intake without adding separate tools.
• Pricing: Quote-based, by users, plan and contract term. No free trial listed; demos available.
• Watch out for: The full AI feature set requires the top plan or paid add-ons.

Filevine (LOIS)

Filevine’s AI, called LOIS, covers drafting in Word, contextual Q&A on case files, deposition summaries and medical chronologies. There is a free tier, LOIS Explore, with document Q&A, cited drafting and legal research, but it does not connect to Filevine case management.

• Best for: Litigation and PI firms that want case management, intake (Lead Docket) and AI on one platform.
• Pricing: All packages are custom-quoted. LOIS Explore is free with no card.
• Watch out for: The free tier doesn’t touch your case data, so the useful features sit behind a paid package.

Personal injury and plaintiff work

EvenUp

EvenUp focuses on personal injury: demand letters, medical summaries, complaints, discovery responses and negotiation prep. It says more than 2,000 PI firms use it. In May 2025 it announced a per-case pricing model covering its drafting suite and workflow tools.

• Best for: PI firms with steady pre-litigation demand volume.
• Pricing: Per case, quoted after a call. No public rate card.
• Watch out for: It is narrow by design. You will still need case management and research tools.

Supio

Supio is built for plaintiff and mass tort firms. It produces medical chronologies, demand letters, case valuation and litigation drafts, and adds firm-level analysis across cases. It lists integrations with Westlaw and case and document management systems, plus an API.

• Best for: Plaintiff firms with large medical record volumes or mass tort dockets.
• Pricing: Demo and quote only.
• Watch out for: Ask how it maps fields into your case management system before you sign.

General AI assistants

General assistants are useful for email drafts, summaries and plain-English client letters. They are not legal research tools. The plan tier decides what happens to your data.

ChatGPT Business

OpenAI’s Business plan (renamed from Team in 2026) costs $20 per user per month billed annually or $25 monthly. OpenAI says Business workspace data is excluded from model training by default, and admins can set retention.

• Best for: General drafting and summarizing across a small team.
• Pricing: $20/user/month annual, $25 monthly; Enterprise by quote.
• Watch out for: Consumer Free and Plus accounts are different. Don’t let staff paste client files into personal accounts.

Claude Team

Anthropic’s Team plan costs $20 per user per month billed annually ($25 monthly) for standard seats, for 2 to 150 users. Anthropic states there is no model training on Team content by default; on Free, Pro and Max, training is something users must opt out of.

• Best for: Long documents, such as summarizing a deposition transcript or a 100-page lease. MyCase and CoCounsel both list connectors into Claude.
• Pricing: $20/user/month annual, $25 monthly; premium seats $100/$125; Enterprise from $20/seat plus usage, 20-seat minimum.
• Watch out for: Same as ChatGPT: business tier only, and no unverified citations.

Microsoft 365 Copilot

If your firm runs on Outlook and Word, Copilot works inside them. Microsoft states prompts and responses stay within the Microsoft 365 service boundary, are not used to train the models, and are available for eDiscovery and retention policies.

• Best for: Firms already on Microsoft 365 Business plans.
• Pricing: Copilot Business add-on $21/user/month annual (discounted to $18 through December 31, 2026 for eligible customers), or bundled with Business Standard at $23.50 and Business Premium at $32.
• Watch out for: Copilot can see whatever a user can see. Clean up SharePoint and Teams permissions first, or it may surface files from the wrong matter.

The part most tool lists skip: making them work together

Each of these tools is good at one job and mostly blind to the others. The intake form doesn’t create the matter in Clio. The research tool doesn’t know what’s in your document system. The demand letter tool can’t see the medical records until someone downloads and uploads them. In a small firm, a paralegal ends up as the glue, copying data between screens.

A few examples of what connecting them looks like. A new lead fills out a web form; an AI agent runs a conflict check against your practice management system, creates the contact and matter, sends the engagement letter for e-signature, and books the consult, with a human approving before anything goes out. That is the core of client intake automation. Or: when medical records arrive by email, they are filed to the right matter, summarized into a chronology, and the deadline calendar is updated. Or: a signed retainer triggers a first-draft engagement packet and standard pleadings built from your own templates, which is what document automation covers.

Where confidentiality is the main worry, there is another option: run the model inside your firm’s own cloud account (AWS Bedrock, Azure or Google Vertex), so client files never leave infrastructure you control and never train a public model. This is often the cleanest answer to the Rule 1.6 question for firms handling sensitive matters.

We build these connections on top of the tools firms already own rather than selling another subscription. If you want to see where the copy-and-paste work is in your firm, the AI audit maps it out. More on how this applies to legal practices is on our law firms page.

Confidentiality and compliance checklist

• Read ABA Formal Opinion 512 (July 29, 2024). It applies the existing rules on competence, confidentiality, communication, supervision, candor and fees to generative AI. It is persuasive, not binding, so pair it with your state’s rules. (ABA)
• Get real consent where it’s required. Opinion 512 says informed consent is needed before putting client information into a self-learning tool, and boilerplate in an engagement letter does not count. (Future Law summary)
• Bill fairly. Under Opinion 512 you can’t bill clients for time spent learning a tool, or for hours you didn’t actually work because AI sped things up. (JD Supra)
• Check your state bar. Florida (Opinion 24-1), Texas (Opinion 705), Oregon, North Carolina, New York, D.C. and others have issued their own guidance, and California updated its practical guidance in 2026. Virginia is taking comment on a proposed opinion about duties when opposing counsel files unchecked AI output. (Florida Bar, NetforLawyers compendium, Virginia Lawyers Weekly)
• Verify every citation, every time. Sanctions keep getting heavier. Mata v. Avianca (S.D.N.Y. 2023) brought a $5,000 sanction for six fake ChatGPT cases. In Noland v. Land of the Free (Cal. Ct. App. Sept. 2025), a lawyer was sanctioned $10,000 and referred to the State Bar. In Johnson v. Dunn (N.D. Ala. July 2025), the court disqualified three lawyers from the case and referred them to the bar, saying fines had not deterred the problem. (Bloomberg Government, McGuireWoods, EDRM)
• Write a short firm AI policy. List approved tools and tiers, ban personal accounts for client work, and require citation checks before filing. Opinion 512’s supervision duties cover staff and vendors too.

Sources

• Thomson Reuters, CoCounsel Legal product page: https://legal.thomsonreuters.com/en/products/cocounsel-legal
• LexisNexis, Lexis+ with Protégé: https://www.lexisnexis.com/en-us/products/lexis-plus-protege.page
• LexisNexis, Lexis+ with Protégé free trial: https://www.lexisnexis.com/en-us/products/lexis-plus-protege/signup.page
• LexisNexis, Lexis+ (small firm packages): https://www.lexisnexis.com/en-us/products/lexis-plus.page
• Spellbook pricing: https://spellbook.com/pricing
• Harvey: https://www.harvey.ai/
• Clio pricing: https://www.clio.com/pricing/
• LawNext, “Clio Work Is Now Available To Solo and Smaller Law Firms As A Standalone Product” (April 2026): https://www.lawnext.com/2026/04/clio-work-clios-ai-workspace-is-now-available-to-solo-and-smaller-law-firms-as-a-standalone-product.html
• MyCase pricing: https://www.mycase.com/pricing/
• Smokeball pricing: https://www.smokeball.com/pricing
• Filevine pricing: https://filevine.com/pricing/
• EvenUp: https://www.evenuplaw.com/
• EvenUp per-case pricing announcement (Business Wire, May 2025): https://business.mammothtimes.com/mammothtimes/article/bizwire-2025-5-15-evenup-launches-ai-drafts-suite-smart-workflows-medical-bill-summary-and-case-based-pricing-to-redefine-legal-automation-for-personal-injury-firms
• Supio: https://www.supio.com/
• OpenAI, ChatGPT pricing: https://openai.com/chatgpt/pricing/
• OpenAI, Managing data, sharing and privacy in ChatGPT Business: https://help.openai.com/en/articles/8798634-managing-data-sharing-and-privacy-in-chatgpt-business
• OpenAI, Enterprise privacy: https://openai.com/enterprise-privacy/
• Anthropic, Claude pricing: https://claude.com/pricing
• Microsoft 365 Copilot pricing: https://www.microsoft.com/en-us/microsoft-365-copilot/pricing
• ABA Formal Opinion 512: https://www.americanbar.org/content/dam/aba/administrative/professional_responsibility/ethics-opinions/aba-formal-opinion-512.pdf
• Future Law (2Civility), “Breaking down the ABA’s guidance on using generative AI”: https://www.2civility.org/breaking-down-the-abas-guidance-on-using-generative-ai-in-legal-practice/
• JD Supra, “American Bar Association Issues Formal Opinion on Use of Generative AI Tools”: https://jdsupra.com/legalnews/american-bar-association-issues-formal-6626731
• The Florida Bar, ethics guidelines for generative AI: https://www.floridabar.org/the-florida-bar-news/board-of-governors-adopts-ethics-guidelines-for-generative-ai-use
• NetforLawyers, ethics opinions on lawyers’ use of generative AI: https://www.netforlawyers.com/content/ethics-opinions-covering-lawyers-use-generative-ai-artificial-intelligence
• Daily Journal, California State Bar updated AI guidance (June 2026): https://dailyjournal-prod.herokuapp.com/articles/391883-new-guidance-proposed-rule-changes-on-generative-ai-in-legal-practice
• Virginia Lawyers Weekly, proposed ethics opinion on AI errors (May 2026): https://valawyersweekly.com/2026/05/29/virginia-state-bar-ethics-opinion-ai-errors/
• Bloomberg Government, Mata v. Avianca $5,000 fine: https://news.bgov.com/esg/chatgpt-phony-legal-filing-case-gets-lawyers-a-5-000-fine
• McGuireWoods, Noland v. Land of the Free $10,000 sanctions: https://www.mcguirewoods.com/client-resources/alerts/2025/9/california-appellate-court-issues-10k-sanctions-in-states-first-published-opinion-on-ai-hallucinated-case-citations
• EDRM, AI sanctions in Johnson v. Dunn: https://edrm.net/2025/07/when-ai-policies-fail-the-ai-sanctions-in-johnson-v-dunn-and-what-they-mean-for-the-profession/