Connect AI

Sussur does the listening on your Mac. The coaching itself is generated by an AI model, and you choose which model and pay for it directly.

Bring your own key#

There is no Sussur account and nothing to subscribe to. You give the app credentials for a provider you control, and it uses them.

• OpenRouter (recommended). Create a key at openrouter.ai, paste it into Sussur's settings, and pick a model. Usage is billed by OpenRouter to you.
• Ollama (offline). Run a model locally with Ollama and point Sussur at it instead. Slower, and only as good as the model you run — but the coaching step never leaves the Mac.

Your key is stored in the macOS Keychain on your machine. Sussur has no server that sees it, and no way to use it.

What actually leaves your Mac#

Worth being precise about, because it is the one network hop in the product:

• Audio and video never leave the device. Capture and speech-to-text run locally.
• Transcript text does leave — directly to the provider your key belongs to — to generate each coaching suggestion and the post-call report. On Ollama, it does not leave at all.
• No facial or vocal signal leaves, because Sussur derives none — it makes no inference from faces, expressions, or tone of voice.

What that provider then does with the text is governed by their terms, not ours. Read them; we are not in a position to promise anything on their behalf.

Full detail in Privacy, Recording & Consent.

One more thing: the speech model#

The first time you run a session, Sussur downloads a one-time on-device speech model (~640 MB). It transcribes your calls locally, with no network. Start it now and it'll be ready by the time you finish setup.