Session chat

Session chat lets you ask questions about one meeting in plain language and get answers grounded in that meeting's own content. Open any meeting page and click Chat.

How it works

When you open chat on a meeting, Hyprcore gives the model that meeting's full transcript and notes as context. You ask a question; the model answers from what's actually in the transcript and notes. If the answer isn't there, it tells you so instead of making something up.

Because the whole meeting is in context, there's no cross-meeting retrieval to configure — the chat sees everything that was said in this recording and nothing from your wider vault.

Good questions to ask

• "What did we decide about pricing?"

• "What action items came out of this call, and who owns them?"

• "Summarize the second half of the meeting in three bullets."

• "Did anyone raise a concern about the timeline?"

Answers are best when the meeting has a good transcript — clean speaker labels and accurate text give the model more to work with.

Picking a model

Session chat uses your post-processing model, configured in Settings → Models. Any of the supported providers work; longer, careful models tend to give better-grounded answers on a dense transcript. If you're signed in, Hyprcore Cloud is the simplest choice; you can also point chat at your own provider or a local model.

Want answers across many meetings — or actions?

Session chat is deliberately scoped to one meeting and only reads. For anything broader, use the agent on the Knowledge page: it reasons across your entire vault, prepares for upcoming meetings, tracks commitments, and — with your approval — can create notes, run sandboxed commands, and remember context between conversations.