Upload the recording. Vecbase listens, writes the minutes, and pulls out every to-do with who owns it and when it's due. Then it sends the summary to Slack, emails the team, and saves a copy — you don't have to open another tab.
Vecbase Agent transcribes the recording, splits it by speaker, pulls out every commitment, and cross-checks it against last meeting's open actions — all in one pass.
| Owner | Action | Due |
|---|---|---|
PSPriya Sundaram | Finalize Identity Federation GA cut · confirm May 22 ship date | 2026-05-13 |
ATAnika Thiele | Brief Latham & Atlas POCs on the manual-migration plan | 2026-05-15 |
JKJordan Kim | Lock Drive sidebar redesign spec · share Nov-14 ship plan | 2026-05-22 |
MVMarcus Vega | Kick off 3 confirmed Q4 experiments · assign DRIs + metrics | 2026-05-16 |
You configure the delivery once. After every meeting the Agent fans the minutes out to Slack, email, Notion, and Linear — so people see them in the tools they already have open.
Drop in an .mp3, .m4a, .wav, or an existing transcript. Multi-speaker audio is automatically diarized — speakers are tagged by voice and matched to attendee names when you provide them.
In the chat: who attended, which channels to deliver to, anything that should stay confidential. Vecbase Agent treats it like a conversation — iterate freely until the minutes read the way your team reads them.
Slack gets the summary, email goes to attendees with the full doc, Notion gets the archived version, and Linear gets one issue per action item. The .docx itself is saved to your Drive so anyone with access can edit it later.
Speaker diarization out of the box — "Priya said…" not "Speaker 3 said…". Technical vocabulary (SLO, NPS, OKR, MAU, ARR) and product code names stay intact. Mixed English-Chinese meetings are handled in one pass.
Not just a transcript dump. The Agent reads the meeting like a chief of staff would — "Priya, can you have the metrics dashboard ready by Friday?" becomes a row in the action items table, assigned to Priya, due 2026-05-15.
The minutes never live in just one place. Configure once and the Agent posts the summary to your team channel, mails the doc, files the archive, and creates issues for every action — so people see it where they already work.
The recording uploads to your workspace bucket and the transcription runs inside an isolated sandbox tied to your account. We do not train models on your conversations. Recordings are kept in your Drive on the retention policy you set.
MP3, M4A, WAV, and FLAC for audio. If you already have a transcript, drop a .txt or a .vtt (Zoom / Google Meet export format) and the Agent skips straight to the minutes.
Sign in, hand it over to the Agent — the finished file lands in your Drive.