Skip to content

Connect OpenAI to atbridge.ai

Prefer to watch? Connect OpenAI to Thunderbird with your API key:

Once OpenAI is connected, the atbridge AI Chat inside Thunderbird uses GPT to read, search, draft, and organise your email — across every account, on your machine.

  1. Sign in at platform.openai.com/api-keys.

  2. Click Create new secret key, give it a name (e.g. “atbridge.ai”), and copy it. It starts with sk-.

  3. OpenAI shows the key once — copy it now; you can’t see it again later.

  1. In Thunderbird, open the atbridge Settings — the gear, or Tools → Add-ons and Themes → atbridge.ai → Options (see Set up the atbridge extension). Go to the AI providers tab.

  2. Pick OpenAI from the Provider dropdown.

  3. Paste your sk-… key into API key. It’s encrypted on this device — your key never reaches atbridge.ai’s servers.

  4. Click Connect. The status turns to a green ✓.

Open the atbridge AI Chat (the @b button in Thunderbird), pick an OpenAI model from the model dropdown — gpt-4o-mini is a fast, low-cost default; gpt-4o is stronger — and ask away:

“Summarise the unread thread from Maya and draft a reply.”

4. Optional — answer from your own knowledge base

Section titled “4. Optional — answer from your own knowledge base”

If you keep a Vector Store in your OpenAI account — a product catalogue, a support handbook, your website’s content — atbridge can look things up in it while it drafts, so replies to customer mail use your real answers instead of guesses.

  1. In the OpenAI dashboard, open Storage → Vector stores and copy the id of the store you want (it looks like vs_abc123…).

  2. Back in atbridge Settings → AI providers, with OpenAI selected, paste it into Vector store ID.

  3. Click anywhere outside the field (or press Enter). It saves itself — a short Saved. appears underneath. Done: no other change to how you work.

From then on, when a reply needs something from that store, OpenAI searches it and uses the matching passages as it writes. It happens inside the same request, so there’s no extra waiting, and it works in the AI Chat and in the compose dock.

Was this page helpful?