Claude for Legal Got a Gallery of Ready-Made Tools. Here Is What a Small Law Office Can Use This Week.
The big firms get the polished version first. The method behind it works in any office: AI does the first pass, every answer points to its page, a lawyer decides.
By John Martines · October 30, 2026 · 6 min
On October 28 Anthropic added ready-made tools for contract review and citation checks to Claude for Legal. A six-lawyer office may not see them yet, but it can borrow the method today. Here is how, in order.
A client drops off nine vendor contracts on a Tuesday. They're buying a small HVAC company and want to know which contracts can be cancelled, which renew on their own, and which ones change if the owner changes. In a four-lawyer office, that's an associate's whole Wednesday: read each contract, fill in a spreadsheet, flag the odd clauses.
Now picture the same job with a table that does the first read. Nine rows, one per contract. Columns for term, renewal, notice period, change of control. Every answer has a link back to the clause it came from. The associate spends Wednesday morning checking the table against the pages and the afternoon on the two contracts that really need a lawyer.
That table exists. On October 28, 2026, Anthropic added it, and a set of tools like it, to Claude for Legal.
Claude for Legal is Anthropic's package of Claude for law firms and in-house legal teams. It launched on May 12, 2026, with practice-area add-ons (Anthropic calls them plugins) and connectors to legal software.
The new piece is a gallery of artifacts. An artifact is a small working tool inside Claude, like a table, a tracker or a checklist, that you load your own documents into and share with colleagues. The gallery includes:
- A contract review table that answers the same questions across a stack of agreements and links each answer to the clause it came from.
- A cite checker that compares a brief or memo against its sources and flags quotes that don't match, wrong page numbers and claims the source doesn't support.
- A tracker for new rules, an intake form for legal requests, and a matter board.
- Tools built by others, including Goodwin, Foley & Lardner, PwC and the legal AI company Harvey.
Anthropic says sharing settings keep teams walled off from each other's matters. The announcement files all of this under Claude Enterprise, its plan for large organizations, and doesn't say whether smaller plans get the gallery. A small office on Claude Team or a personal plan may not see it yet.
How it got here:
- June 22, 2023: A federal judge in New York fined two lawyers and their firm $5,000 in Mata v. Avianca for filing a brief with cases ChatGPT made up.
- July 29, 2024: The American Bar Association issued Formal Opinion 512, its first ethics guidance on generative AI, covering competence, client confidentiality and fees.
- February 2026: Anthropic released a legal plugin for Claude.
- May 12, 2026: Anthropic launched Claude for Legal, with 12 practice-area plugins and connectors to legal software.
- October 28, 2026: Anthropic added the gallery of ready-made legal artifacts.
If you run a small law office, the gallery itself may be months away. The shape of it is available now, and it answers the question most small firms have been stuck on: how do we use AI without ending up like the lawyers in Mata?
Three things in the design are worth copying.
Every answer points to its page. The contract table and the cite checker don't ask you to trust a summary. They show where the answer came from, so checking is fast. That's the difference between a tool that saves an hour and one that creates a malpractice risk.
The job is narrow. "Find the renewal terms in these nine contracts" is a job AI does well. "Advise the client on the deal" is not. Each tool in the gallery does one repeatable task.
A person signs off. Anthropic presents these as a first pass, and the ABA's opinion says the same thing in ethics terms: you stay responsible for the work, so you check it.
If you don't run a law office, this still touches you. Your lawyer may start doing contract reviews faster, which should show up in the bill. Under the ABA opinion, a lawyer may bill for time actually spent, not for time the AI saved. It's a fair question to ask at your next engagement.
1. Pick one repeatable job. Good first choices: pulling dates and terms from a stack of leases or vendor contracts, comparing a new draft against your standard form, or listing every deadline in a scheduling order. Not legal research for a filing, not yet.
2. Use a business account, never a personal one. Client documents go only into an AI plan the firm pays for, whose terms say your data isn't used to train models. Read those terms once and keep a copy in your office file. The ABA opinion expects you to understand the tool's terms, and in some cases to get the client's informed consent.
3. Build the table yourself. You don't need the gallery to copy the method. Attach the contracts to a new chat and ask for a table with one row per contract, the columns you care about, and the page or section number for each answer, with "not found" where a term is missing.
4. Check every row against the page. The table tells you where to look. A lawyer or paralegal opens each cited page and confirms it. Mark each row as checked. That habit is the whole safety system.
5. Never file a citation you haven't read. If AI suggests a case, pull it from Westlaw, Lexis, your state's free research service or the court's own site, and read it before it goes in anything you sign.
6. Write it down. One page: which AI tool the office uses, for which jobs, who checks the work, and what never goes in. Our free [Staff AI Rules Kit](/tools/staff-rules) builds that page from your answers in a few minutes.
- Is this a firm-paid account whose terms rule out training on your data?
- Does every answer come with a page, clause or quote you can check?
- Who checks the output, and is that written on the matter?
- Would the client be surprised to learn AI did the first pass? If yes, tell them first.
- Will your time entries show the time actually spent, not the time the job used to take?
Legal AI is moving from a chat box to small, shared tools built for one task, with a trail back to the source. Anthropic says more are coming from other firms and legal software makers, and the tools will likely reach smaller plans in time.
The firms that do well with this won't be the first to get the gallery. They'll be the ones that already know their one repeatable job, already check every row against the page, and already have the rule written down. Start with nine contracts and a table this week.
---
Sources (checked October 30, 2026). Gallery of legal artifacts, contract review, cite checker, sharing settings and ethical walls, partner-built tools: Anthropic, "Introducing artifacts for legal work" (claude.com blog, Oct 28, 2026) and Claude on X (Oct 28, 2026). Coverage, including the Enterprise filing and no word on other plans: "Anthropic Increases Legal Vertical Push With 'Artifacts'" (Artificial Lawyer, Oct 28, 2026); "Claude's new legal artifacts put ready-made tools in front of your whole firm" (Intelligence by Intent, Oct 2026); Crypto Briefing and YourStory (Oct 2026). Claude for Legal launch, plugins and connectors, February legal plugin: "Anthropic launches Claude for Legal" (ABA Journal, May 2026); Legal Technology Hub and Global Legal Post (May 2026). Mata v. Avianca: "Court imposes sanctions on lawyers who filed bogus cases after relying on ChatGPT" (LawNext, June 2023); Bloomberg Law (June 22, 2023). ABA Formal Opinion 512: "ABA issues first ethics guidance on a lawyer's use of AI tools" (ABA News, July 29, 2024); The Florida Bar News (2024).