The meeting doesn't end when the call does
Zoom, Google Meet, Teams — meetings are where a lot of real thinking happens and almost none of it gets captured. Here's how tentus closes that gap without asking you to do anything different.
A striking amount of the thinking that shapes a week happens in meetings, and almost none of it survives the meeting. Someone takes rough notes, or nobody does, and the decisions and commitments that got made live only in memory until they quietly evaporate.
The invisible-by-default approach
tentus connects natively to Zoom, Google Meet, and Microsoft Teams. When a meeting you're the host of gets recorded, the transcript lands in your workspace automatically — no bot joins the call, no separate app to remember to open, nothing to configure per meeting. It's the same principle as everything else we build: the fewer things you have to remember to do, the more likely the thing actually gets done.
When invisible isn't an option
Not every meeting is one you're hosting, and not every platform has a webhook that tells us the moment a recording finishes. For those cases — and for Webex, which none of the invisible integrations cover — there's Vexa, a visible notetaker bot you can add to a call the way you'd add Otter or Fireflies. It's the deliberate trade-off: less invisible, more universal. Different meetings call for different levels of ceremony, and we'd rather offer both than force one model on every call you take.
What happens after the transcript lands
This is the part that matters more than the capture itself. A transcript that just sits in a folder is barely better than no transcript. Inside tentus, it becomes a source like any other note — indexed, embedded, connected to whatever else in your workspace shares its entities or its themes. If a commitment inside that meeting sounds like "I'll follow up with Sarah by Friday," Lumi's proactive layer can pick that up and offer to turn it into a task, one confirm click away.
Why we didn't build our own transcription model
Zoom, Google, and Microsoft already do a competent job of turning audio into text at the platform level, and Vexa (self-hosted, routed through a cheap Whisper-class model) covers the rest. Building a proprietary transcription engine would have been effort spent duplicating something already solved, instead of effort spent on what happens to the transcript once it exists — which is where tentus actually adds something new.