- The server joins a take to the answer that drew its proposal; the card only
looked in later answers, so it never found the take to decide and "Plan
this" still waited below for a second confirmation.
- An answer that asks the user something offers no fixed chips at all, only
the model's own answers to its question.
- The model proposes a clear request at once, saying what it assumed, and
puts the way it recommends first, which the card shows chosen.
A turn may end on suggest_follow_ups: the server answers the call itself, keeps
its questions as data-trek-follow-ups and the chat shows them as chips under
the answer. Without them the chips fall back to what the answer holds, now by
the kind of change: a tour asks about the tour, the day plan's questions only
follow a change to the day plan, and an answer that asks the user something
offers no "tell me more".
Plan this checks place_edit like the tour routes, the include choices of a
run or group stay disabled while the chat applies a change, and a follow-up
can say when it applies, so Confirm the pending bookings shows only when the
card holds a pending booking of this trip.
The three AI addons share one tile in the integration column; a dialog holds
the connection, the assistant's limits and AI parsing, which now reads with
the assistant's connection unless an install keeps its own. Existing parsing
configs stay where they are and keep winning.
The chat sees the map on screen, offers next questions under an answer,
shows its live step in the window head, previews waiting changes inside the
day list, checks whether a day still works, folds a multi-day plan into a
week grid, keeps per-trip notes and can estimate costs. An empty or cut-off
answer gets one push instead of ending silently, and the assistant's own
step and size limits are raised.
Maurice