AI for legal work, with the rules around it
Claude, Anthropic's AI assistant, set up for a law firm or legal team on company accounts, with a policy on what goes into it and rules on who checks what comes out. AgiLawyer is a member of the Claude Partner Network.
What it is
Claude is the AI assistant built by Anthropic. Lawyers use it to draft, review and summarise documents, compare versions and work through research questions, with the documents of a matter kept together in one project.
On Anthropic's business plans the firm owns the accounts and decides who has access, and Anthropic's commercial terms do not allow it to train models on the firm's content. That makes Claude usable under a duty of confidentiality, once the firm has set its own rules.
We use Claude in our own work, so we recommend it from daily use, not from a demo, and we say so where another tool fits you better.
What AgiLawyer configures
- Company accounts in place of private ones, with access managed by the firm
- An AI policy on one page: what may go into Claude, and what may not
- Review rules: what a lawyer checks, when, and what gets recorded
- Projects per practice area, with your templates, precedents and style guide
- Connections to the systems you already use, where a connector exists
- Hands-on training by role, on your own anonymised documents
Claude by Anthropic is the right fit when
- Your lawyers already use AI on private accounts, and you want that use visible and allowed
- You draft, review and summarise long documents every day
- You need rules you can show a client or a regulator
- You want to start with a few tasks, measure the time saved, then decide
Data, security and terms
As the vendor states them, with the page each comes from. Check them against the version you sign.
- Training on your content
- Anthropic may not train models on customer content from its commercial services.
- Source: anthropic.com
- Ownership and confidentiality
- The customer owns the outputs, and its content is treated as its confidential information.
- Source: anthropic.com
- Certifications
- SOC 2 Type I and Type II, ISO 27001:2022, and ISO/IEC 42001:2023 for AI management systems.
- Source: support.claude.com
- Business plans
- Team: single sign-on, projects and connectors. Enterprise adds user provisioning (SCIM), audit logs and your own data retention period.
- Source: claude.com
Questions we are asked
Is it safe to put client documents into Claude?
On a business plan, Anthropic's terms bar it from training on your content and treat it as your confidential information. Which documents may go in is still your firm's decision, and the AI policy records it. Private accounts fall under consumer terms instead, which is one reason to move use onto company accounts.
Team or Enterprise?
Team covers single sign-on, projects and connectors. Enterprise adds user provisioning, audit logs and your own retention period, which matter when a client or regulator asks who did what. We recommend one once we know how many people will use it and what you must be able to show.
We already have Microsoft Copilot. Why Claude?
You may not need it. We train teams on Copilot, ChatGPT, Claude, Gemini and CODEXIS AI, and recommend the tool that fits the tasks. Running both is possible: Copilot inside Microsoft 365, Claude for work on long documents.
Can we start small?
Yes. We pick a few tasks that repeat, measure how long they take today, set up Claude for them and compare. You decide on the rest with the numbers in hand.
What does it cost?
Anthropic publishes the price of each plan on claude.com/pricing. What our setup costs depends on the number of users and tasks; we quote it after a call.
Does the partnership make you less independent?
Membership of the Claude Partner Network gives us training and technical support from Anthropic. We still recommend against Claude where another tool fits better, as we do for every partner on this page.
See Claude by Anthropic with your own matters
We'll show how it would work for your team, and tell you if another tool fits better.