Stenograf help

Moving from another notetaker

The meetings you already have can come with you. Each one becomes a saved recording in History, with its speakers and times, and Discover searches it like anything recorded here. Granola, Otter and Fireflies can be connected so everything comes over at once and keeps coming; the others are a file or a paste, and nothing leaves your computer.

Connect Granola, Otter or Fireflies

In Settings → Integrations, press Connect on the Granola, Otter.ai or Fireflies.ai row and sign in in your browser. Stenograf then asks the tool’s own server — the same one its AI integrations use — for your meetings and their transcripts, saves each as a recording, and checks for new ones every fifteen minutes while it is running; Sync now asks sooner. It sends the tool only your sign-in and the identifiers of meetings; nothing recorded here goes to Granola, Otter or Fireflies, and Soniqo sees none of it. Disconnect forgets the sign-in on this computer and keeps what was imported; the tool keeps Stenograf authorised until you remove it in its own settings.

What a tool shares depends on your plan with it: Granola shares transcripts on its paid plans only and lists just the last thirty days on the free one, so a free account sees its meetings counted as “without a transcript” rather than imported. Otter’s help center says which of its plans include its MCP server.

Where the file import is

In History, choose Import transcript. Say which app the meeting came from, then either pick the file in the system dialog or switch to Pasted text, give the meeting a name and its date, and paste. One meeting at a time; the same export is never imported twice. While History has fewer than three recordings a card offers the same import, and Not now puts it away for good.

Otter.ai

Connecting Otter, above, brings everything over. For one conversation, open it, choose Export, and take TXT with speaker names and timestamps switched on — every Otter plan has it. Paid plans also offer SRT, which imports the same way. Pick Otter.ai in the dialog and choose the file.

Granola

Connecting Granola, above, is the easy route. By hand, Granola has no per-meeting file to download: open the meeting, switch from the notes to its transcript, and use the copy icon in that pane. In Stenograf pick Granola — the dialog opens on the text box — then name the meeting, set its date and paste. Granola writes Me for your side and Them for everybody else; Stenograf keeps that split the way a recording made here has it, your words on the microphone side and the rest as the meeting.

Paste the transcript, not the notes. Granola’s notes are its own model’s summary; Stenograf writes a summary of its own from the transcript once it is imported. Granola’s once-a-day CSV export holds titles and summaries, not transcripts.

Fireflies.ai

Connecting Fireflies, above, brings everything over, speakers and times included. By hand, open the meeting, choose Download on the transcript, and take JSON: it carries every sentence with its speaker and times, and Stenograf joins the sentences back into paragraphs. Downloads are available on Fireflies’ Pro plan and up. TXT, MD and SRT import too, with the speaker names and timestamps options switched on. The Moving from Fireflies guide walks through it step by step, including stopping the Fireflies bot afterwards.

Zoom

A cloud recording has an Audio transcript beside it, a .vtt file with each participant’s name in front of what they said; pick Zoom and choose it. A local recording is audio rather than a transcript: use Import audio in History instead, which transcribes it here with speakers matched against your saved voices, on macOS for now.

Anything else

A .txt, .md, .srt, .vtt or .json from any other tool, or text you paste, is read the same way: paragraphs, Name: words lines, caption cues, or a JSON list of sentences under the usual keys. Pick Another app.

What carries over, and what stays behind

Speakers and times. Names come across exactly as the other tool wrote them, as labels on the transcript. They never become voice profiles: no audio was heard, so there is nothing to recognise a voice from. Where the file gives no times, paragraphs are spaced by their length so the transcript still reads in order.
Name and date. A file gives its name and its date; pasted text takes the ones you type, so a meeting imported months later still belongs to the day it happened.
Where it came from. History says Imported from Otter.ai, and so does Discover when it cites the transcript.
Everything built on the transcript. A summary in the template you use, History search, Discover, and read-only access for an AI assistant you have authorised, exactly as for a recording made here.
Not the audio, and not the other tool’s notes. Audio you still have goes through Import audio; notes and summaries another model wrote are not imported, because Stenograf writes its own from what was said.

Nothing about an imported meeting leaves your computer. Stenograf does not sign in to Otter, Granola, Fireflies or Zoom, reads no account and no local cache, and sends nothing anywhere.