The recap email, verbatim.

Three to four sentences capturing the meeting's shape — who, what, why it mattered. Reads like the lede of a news article, capped at 400 characters. The first thing your eye lands on.

Up to five top-level takeaways the AI pulled from the conversation. Distinct from action items — these are statements about what happened in the room, not commitments anyone made.

Every commitment from the call, rendered as ☐ checkboxes with the assignee in bold. Up to eight items. Paste straight into your CRM, project tool, or weekly. Each item is one decision someone owns.

Per-speaker analytics — speaking time and percentage, turn count, longest single run, plus qualitative counts (questions, statements, commitments). Talk-time bar makes the split obvious at a glance; the qualitative counts add context the bar alone can't show (a 50/50 split where one side asked all the questions reads very differently from a 50/50 where both sides made statements).

Places where someone softened a commitment or signalled uncertainty. The classic 'I think', 'maybe', 'we'll try to' moments. Surfaced under Next steps because the framing is intentional: 'you said you'd try — make it concrete or let it go'.

Items that came up but were explicitly deferred to a future meeting or DM. Tagged with the reason if one was given. Means nothing falls through the cracks — the punt itself is recorded with context.

Calendar agenda items the meeting didn't reach. The 'we said we'd talk about pricing but we didn't' list. Catches dropped balls before they cost you on the next call.

Click the link at the top of the recap email to open the meeting in your dashboard and ask anything — 'what did Aisha commit to?', 'when did pricing come up?', 'what's their evaluation timeline?'. Answers come back with the exact transcript line they came from, so nothing is hallucinated. Works across a single meeting on every plan, and across your whole meeting history on Pro+.