Add the server in seconds
Drop the White Shoe MCP server into your AI client's config. One entry. Thirty seconds — by an admin or the lawyer themselves.
Your Firm IQ, playbooks, and matter repository — exposed to Claude, ChatGPT, and any MCP-compatible client. Counsel-grade context, wherever the thinking is already happening.
Claude. ChatGPT. Any MCP-compatible client.
Three movements. No new portal. The model already knows how to call a tool — we hand it the right ones.
Drop the White Shoe MCP server into your AI client's config. One entry. Thirty seconds — by an admin or the lawyer themselves.
Authenticate with your workspace. Scoped OAuth tokens. The model host never sees your firm data — only the answers you authorize the server to return.
Ask in Claude. Ask in ChatGPT. The model now answers with your playbooks, your matters, your firm.
Ask the question in the AI you already use. The Associate handles the rest, in the client where the lawyer already is.
The MCP server exposes the same intelligence that drives every White Shoe surface — packaged as tools and resources the model can call by name.
Preferences, voice, and house rules — exposed as queryable context to any MCP client. The model stops guessing.
Negotiation positions, fallback language, and risk thresholds — looked up in plain language by the model, returned in the structure your team already uses.
Semantic search across your matter archive. Source documents returned in-chat, with citations the lawyer can verify.
Call any Associate as a tool. Redline. Triage. Draft. The same workflows that drive the platform — available wherever the model can call a tool.
The team already knows Claude. They already use ChatGPT. The legal context should show up where the thinking is — not behind a new login.
The team already lives in their AI client. The firm's brain shows up there too — same model, same UI, more context.
MCP gives the model access to Firm IQ, your playbooks, and your repository — so the answer isn't generic, it's yours.
Scoped OAuth tokens. The model host is never trained on your data. Encrypted in transit. The model sees what you authorize — nothing more.
Bring your own model.
Bring your own firm.
Coming Q3 2026. Available on every plan. Join the waitlist for early access — we're rolling out a private beta first to existing customers.