Claude for lawyers and law firms: a practitioner's guide

Claude can take real drafting and reading off a lawyer's desk, as long as you decide what goes into it. Here is what it costs a solo practitioner or a small firm, the settings to check before anyone pastes client material, prompts that work, and the part of client work it was never built to hold.

This guide is for lawyers and small law firms deciding how to use Claude in client work: the solo practitioner, the three-partner firm, the attorney who already has it open in another tab. It sits under our practitioner's map of AI for lawyers, which covers the wider landscape: legal research platforms, bar guidance and firm policy. Here the subject is one tool, read at Anthropic's own pages on 2026-10-12, from the price to the privacy switch. Anthropic's legal release, Claude for Legal, gets two sentences here and a page of its own.

One disclosure first. This guide is published by Equerry, a personal assistant for the client side of practice. Equerry is not a legal research or drafting tool, and it is not affiliated with Anthropic. Nothing on this page is legal advice. Whether client material goes into any tool is a question for your firm's policy, your jurisdiction's guidance and, where it applies, your client.

What Claude is, and what it costs a lawyer or a small firm

Claude is Anthropic's general-purpose assistant, on the web, the desktop and your phone. It drafts, rewords, summarises and reasons over whatever you give it. Out of the box it is not a legal research platform, and anything it tells you about the law is unverified until you check it at the source.

Every plan has usage limits. On Pro, Anthropic's help center says the session limit "will reset every five hours," with a weekly limit on top. For most solo practitioners the realistic path is Free to test, then Pro. The difference that matters most for client work isn't usage, though. It's training: on Free, Pro and Max, Anthropic's plan comparison lists model training as "Opt-out," which means it is a setting you control (the steps are below). On Team it reads "No model training on your content by default."

For a firm rather than a solo practice, the choice is usually Team against a handful of individual Pro accounts. Team brings central billing, single sign-on and the no-training default with it; separate Pro accounts leave the training switch, and every other setting on this page, in each lawyer's own hands. Either way, the firm's policy on what may be pasted comes before the plan.

Claude for Legal, in two sentences

On May 12, 2026, Anthropic published "Claude for the legal industry," introducing, in its words, "20+ new MCP connectors that link Claude to the software the legal industry already relies on, and 12 new plugins tailored to specific legal work and practice areas," which the post says are "open source and available in Claude Cowork." What the connectors reach, what each practice-area plugin does, who it is built for and how a small firm can use it are on our page Claude for Legal.

Where Claude saves a lawyer time, and where it falls short

The best uses are the writing and reading that surround the law, not the law itself:

• The prose around the legal work. A status update a client can follow, a cover note, a plain-English account of what happens at a hearing. Claude drafts, and you edit.
• Reading long material you are free to share. A published opinion, a statute, a regulator's guidance; check every passage it relies on against the text.
• Structure. Turning your own rough notes into an outline, a timeline or a list of open items.

Where it falls short is just as predictable. Authority: citations and statements of law have to be checked, which is why Anthropic's own legal plugins treat every output as a draft for attorney review. The matter: Claude knows your matter only through what you have typed, pasted, uploaded or connected; a chat is not the file. The record: its memory holds distilled topics rather than the correspondence, and Anthropic notes: "When a conversation expires or is deleted, related memory entries generated from it won't be removed, but you can delete individual memories at any time."

What to paste, what never to paste, and Claude's settings step by step

A conservative starting point, until your firm's policy says otherwise:

• Paste freely: public material (opinions, statutes, published guidance), your own writing with no client identifiers, templates the firm owns, and general questions.
• Hold back until your policy and, where it applies, your client say yes: client names alongside matter facts, privileged communications, documents produced under a protective order or an NDA, and anything your engagement letter rules out.
• Remember that removing a name doesn't remove the client. A date, a deal size and an industry together can identify one as surely as a name.

Then set Claude up. Five controls, all from Anthropic's help and privacy centers, read 2026-09-29 and again on 2026-10-12:

1. Decide the training setting. On the web or the mobile app: select your name in the settings menu, choose Settings, then Privacy. Anthropic's Privacy Center (article dated August 3, 2026) says: "Under Help Improve our AI models, click the button to toggle it off/on." Other Anthropic pages call the same control "Model Improvement," so look for either name. Anthropic: "If you decide to turn off the model training setting, we will not use any new chats and coding sessions you have with Claude for future model training." If you leave it on, Anthropic "may retain your data in a de-identified format for up to 5 years." Note too that giving a thumbs up or down stores "the entire related conversation" for up to 5 years.
2. Use an incognito chat for one-off work. On a new chat outside a Project, select the ghost icon. Incognito chats are "temporary conversations that aren't saved to your chat history or to Claude's memory," they "are not used for training," and they are still "retained for either 30 days (default), or longer" where an Enterprise organization sets its own retention. Incognito is only available outside Projects, and Anthropic adds: "If you have the new Claude experience, incognito chats open in the previous chat experience, so Claude can't create files or run code in them."
3. Read and prune memory. Open Settings → Memory. "Everything Claude remembers is listed under Topics. Select any topic to read it, then use the edit icon to change it or select 'Delete' to remove it." From the same screen you can switch off Generate memory from chats and choose to pause it or reset it; a reset "permanently deletes all memories including project memories."
4. Keep each matter in its own Project. "Each project has its own separate memory space and dedicated project summary," so one Project per matter keeps each matter's memory separate.
5. Delete what you no longer need. A deleted chat is "Removed from your chat history immediately" and "Deleted from our back-end storage systems within 30 days." Delete the memory topics that came from it too, since deleting the chat does not remove them.

None of these settings answers the confidentiality question for you. They tell you what Anthropic does with what you type. Whether that is enough for a given client is a matter for your jurisdiction's guidance and your firm's policy, which the map linked at the top of this page walks through.

Prompts for client work

Each of these works with the details removed or with material you are cleared to share. Read every word before anything leaves your desk.

• The status update. "Rewrite this status update for a client who is not a lawyer. Keep it under 150 words, lead with what changed and what I need from them, and keep every date exactly as written: [your draft, names removed]."
• The plain-English explainer. "Explain what happens at a deposition to a first-time witness in five short paragraphs. Where something varies by jurisdiction, say so rather than stating a rule."
• The public-document summary. "Summarise this published opinion in 200 words: the question, the holding, the reasoning, and what it leaves open. Quote each passage you rely on so I can check it against the text: [the opinion]."
• The second reader. "Here is the outline of an argument with client details removed. List the three strongest objections the other side could raise and which of my points each one attacks."
• The follow-up after a call. "Turn these rough notes from a call into a follow-up email: what we decided, what each side owes and by when. Flag anything ambiguous in my notes instead of guessing: [notes, names removed]."

The last prompt only works if the notes exist.

What Claude does not hold: what each client said across email and calls

Anthropic's help center says it plainly: "Claude can generate memory based on your chats." That makes Claude good at remembering how you like an update written, not a record of the client. Tuesday's call reaches Claude only if you type it in; the reply from opposing counsel, only if you paste it or connect the system it sits in. Anthropic's connectors "bring your legal work (the documents, communications, and records tied to specific matters) into Claude," which means connecting a whole system, not the one thread you chose. How far any tool may reach is the decision the bar guidance in our practitioner's map of legal AI asks you to make carefully.

You can keep the record with no AI at all: a four-line note within five minutes of the call, a two-line recap email to the client, filed with the matter the same day. The weak point is finding it four months later.

Where Equerry fits

Equerry is a personal assistant who helps you get things done. For a lawyer that means the client record, not the law: it does not research, draft pleadings or review contracts, and it is not a practice-management system and doesn't connect to one. In one matter, it looks like this:

• CC or forward the one thread that matters. Equerry has an address of its own at mail.getequerry.app. Forward it a thread, or add it to the CC line of a live one, and while the address stays on the thread, later replies arrive on their own. It never asks for access to your inbox, and the original stays in your own mailbox. What Equerry remembered from it is listed in Settings → Your memory, each line marked as from that email, and you can delete any line.
• Debrief the call by voice. Walking out of a client call, record a short voice note in Equerry: what was decided, what the client said, what you owe. It is transcribed and remembered. Equerry does not join, record or place calls; the note is yours, made after the call.
• Save what the client sends. Share a PDF, a photo of a marked-up page or a screenshot from your iPhone's share sheet.
• Type a note. When an answer draws on a note you saved, it carries a source chip: tap it to see the passage Equerry used and open the original.
• Ask later, in plain language. "What did the client say about the hearing date?" is answered from what you gave it.
• Walk into the meeting briefed. If you allow calendar access and set up Meeting Prep, Equerry puts together where things stand before a client meeting from what you have shared: the latest reply, what is open, what was agreed last time.

The honest limits: Equerry only knows what you send it, plus your calendar if you allow it, so it cannot find the email you never forwarded. Anyone on a thread can see its address on the CC line. And sending something is still sharing it: what you send is processed to make it searchable and answerable, including by the outside providers named on our AI disclosure page. Equerry makes no claim about privilege or about meeting any bar rule, and whether client material goes into it is a decision for you under your firm's policy and your jurisdiction's guidance, the same as for Claude. What changes is scope: you decide thread by thread and note by note, instead of connecting a whole mailbox.

Keep Claude for the drafting and the reading; it is good at both. For what Claude can see in your mail, read can Claude read my emails?

Sources, first read 2026-09-29 and read again 2026-10-12. Anthropic, Claude for the legal industry (May 12, 2026). anthropics/claude-for-legal README (Apache License 2.0). Claude pricing (prices exclude applicable tax; "Price and plans are subject to change at Anthropic's discretion"). Claude Help Center: What is the Pro plan?, Use Claude's chat search and memory to build on previous context, Use incognito chats. Anthropic Privacy Center: How do I change my model improvement privacy settings? (dated August 3, 2026), Is my data used for model training?, How long do you store my data? Anthropic's pages describe consumer plans (Free, Pro, Max) unless they say otherwise. Vendors change their products and pages often; where Anthropic's current page disagrees with this one, Anthropic's page is right. This guide is general information, not legal advice.

Last updated 2026-10-12.

All product names, logos, and brands are property of their respective owners. Equerry is not affiliated with, endorsed by, or sponsored by any other product mentioned; comparisons are provided for informational purposes.