Ask: chat with everything your meetings have taught it

The agent in your meetings and the Ask chat in your workspace are the same brain at different hours. What one learns, the other can answer.

What Ask is for

Ask is the between-meetings interface: a chat that answers from your meeting history and knowledge base. "What did we promise the pilot customers?" "Who owns the billing migration, and what's blocked?" "When did we last talk to Acme and what changed?" The answers come with the receipts — which meeting, which page — so you can check them rather than trust them.

It answers from your record

This matters enough to say precisely: Ask is grounded in your workspace's own meetings and pages. When it cites, it cites your transcript, your decision page. It is deliberately built not to pad answers with plausible-sounding filler from nowhere — an answer it cannot ground is worth less than "I don't have that," and it knows it.

The retrieval is literally the same

The search that finds context for a chat answer is the same retrieval the agent performs mid-call — one implementation, used twice. This is why gardening the knowledge base pays double: a clearer page improves both the spoken answer in Tuesday's meeting and the written one on Thursday afternoon.

What a message costs

Chat is metered in messages per month — a message is one question, however much retrieval answering it takes, so the counter never moves by an amount you could not predict. Every plan includes an allowance (even the free tier carries a small one, so you can interrogate your free meeting), and message packs exist for heavy months.

Where it shines

The honest use case is the question you would otherwise ask a colleague who then has to go and look: pre-meeting catch-up, "what did I miss last week," writing the status update from what was actually said rather than from memory. The meetings already contained the answers. Ask is just the first tool that makes retrieving them cheaper than re-asking the room.