Skip to content

FAQ

Jump to bottom
Lucas Skog edited this page Jul 10, 2026 · 2 revisions

FAQ

Does it cost anything?

The app is free to download and use. If you run the AI locally or against your company's own server, there are no extra costs. If you use a cloud service such as OpenAI, Azure OpenAI or Anthropic, you pay that provider's usual usage fee directly to them — referat doesn't add anything on top and has no subscription.

Where are recordings stored?

Locally, under:

%APPDATA%\referat\meetings

There is one folder per meeting containing the audio, transcript and minutes. Nothing is uploaded automatically — outbound traffic goes only to endpoints you chose yourself: text to the AI services for transcription and summarization, and (only if you enable the optional speaker identification) audio to a local server you run. See Architecture for the exact file layout.

Which AI services work?

Any OpenAI-compatible endpoint for transcription and chat, plus Anthropic for the minutes. That covers:

  • Local servers such as speaches (transcription) and Ollama (minutes).
  • Your company's internal endpoint.
  • OpenAI and Azure OpenAI.
  • Anthropic (for the minutes).

Transcription and summarization are configured separately. See Configuration.

Does it work with Teams, Zoom and Meet?

Yes. referat records your computer's system audio, so it works with any meeting tool — Microsoft Teams, Zoom, Google Meet — or an in-person meeting in the room. No bot joins the call and no plugin or integration is required. It records what your computer plays plus your microphone.

What language is the app in?

The interface is currently Swedish, and the default minutes template produces Swedish minutes. The transcription follows the spoken language. An English UI is on the Roadmap. This wiki and the landing page are in English. You can also edit the minutes template to output another language — see Configuration.

Is there any telemetry or tracking?

No. The app doesn't phone home. Outbound network traffic goes only to the endpoints you configure yourself — transcription, summarization and, if you enable it, the local speaker identification server. There is no analytics, no account and no usage reporting.

Are my API keys safe?

Yes. Keys are encrypted with Windows DPAPI (via Electron safeStorage) and stored as ciphertext. The plaintext key is never written to disk and never sent back to the app's interface. If OS encryption isn't available, referat refuses to store the key rather than saving it in plaintext.

How are long meetings handled?

referat records in ~10-minute segments. Each segment is an independently decodable file, kept under provider size limits (OpenAI rejects files over 25 MB). At transcription time the segments are processed in order and stitched back together with correct timestamps, so meeting length isn't a problem.

What happens if transcription or summarization fails?

The meeting is marked Error with a plain-language message (for example "wrong or missing API key" or "the server isn't responding"). Nothing is lost — press Try again and the pipeline resumes from the failed step: if the transcript already exists it only re-runs the summarization. If the app is closed mid-processing, the job resumes automatically on the next launch.

Can I close the app while it's working?

Yes. Transcription and summarization run in the background and are saved to disk as they progress. If you close the app, the work resumes the next time you open it.

Related pages

Clone this wiki locally