How Bertram works
From your mailbox to the AI and back, in plain terms.
When you open Bertram, it fetches your messages directly from Gmail or Outlook in real time using the access you granted. It does not keep a permanent copy of your mailbox. Lightweight metadata — things like which threads exist and their priority — is cached briefly so the inbox feels fast, but the full content of your email is not stored permanently.
Where the intelligence comes from
Bertram uses Anthropic's Claude models for the AI work: summarizing threads, drafting replies, and classifying priority. When you ask for a summary or a draft, the relevant message content is sent to the model, the result comes back, and Bertram shows it to you. Your email is never used to train AI models.
The pieces under the hood
A web app handles everything you see and interact with. A backend service talks to your email provider and to Claude, and coordinates summaries, drafts, prioritization, and scheduling. Sign-in is handled by a dedicated authentication provider so your Bertram account is kept separate from your mailbox credentials.