Know every client like they're your only client.

With the connector, client files never reach Briefly's servers. They stay in a folder on your own computer, and only your sign-in, a seat check, de-identified preference rules, and a short usage record for each tool call reach us instead. Firms on the web app keep files in an environment we secure and control, and the firm decides who signs in. Our security page breaks down the full data path if you want more detail.

We don't use client data to train AI models, and any pattern-level learning runs through a de-identified layer with no client name or account number attached. Claude reads the file under your firm's own Claude plan when it answers a question, and for Claude for Work, Claude Enterprise, and the API, Anthropic's commercial terms bar training on customer content. We recommend running Briefly on a business Claude plan for that reason, and our security page covers the difference in full.

When you ask a question, Claude reads the file under your firm's own Claude plan to answer it, and Briefly never stores what Claude reads. On business plans, Anthropic's commercial terms bar training on customer content, and that commitment sits in a contract your compliance officer can review. Our security page lays out the full data path, including how personal Claude plans work differently.

Pasting client details into a personal AI chat window sends them to the vendor's servers under consumer terms your firm never reviewed. The SEC's amended Reg S-P rule requires firms to oversee the service providers that handle customer information, which is why compliance teams are asking about AI chat tools. With the Briefly connector, client files stay in your firm's own folders and Claude reads them under the Claude plan your firm chose and can review like any other vendor.

AI models change every few months, and what a chat window learned about your firm doesn't carry over. Briefly organizes your firm's knowledge into plain files you own, so the context you build keeps working as models improve or get swapped out. Answers draw on your whole book, corrections hold, and every claim carries a citation you can open.

Yes, and you supply everything it works from. Answer five questions about their goals and situation, then attach whatever statements they've shared. Briefly runs up to nine checks against what's actually in those documents, the allocation, the all-in fee, and concentration among them, and hands back findings it can prove alongside questions worth asking at the next meeting. It doesn't make recommendations, and every analysis ends with a plain list of what it could not verify. Read a full sample.

If your team already works in Claude, start with the connector. It reads folders on your own machines and answers where you already work, usually the same day you install it. If you want the file built for you, take the web app instead. We connect Salesforce and Outlook and build the client file, which takes longer on our side, and your team signs in from a browser. A seat can move between them without rebuilding anything.

With the connector there is nothing to migrate, because the files were always in your own folders. Removing it just stops Claude from reading them. On the web app, your firm has ninety days after leaving to export its files in a standard format, and after that window we delete them.