Skip to content
New: simulate any flow against real records before it goes live. See what's new

Give your chatbot an agenda: capture details and qualify leads

Last updated 4 August 2026

An agenda is what a bot is trying to achieve while it answers — gather these details, qualify against these rules, then do this. It is different from a conversation flow: a flow is a script the visitor is walked through, an agenda is a goal the bot pursues in the background of a normal conversation.

Creating one

Go to Chat → Agendas, pick a preset and give it a name:

  • Capture lead — name, email, phone into a contact.
  • Pre-qualify a lead — the above plus company, team size, budget and timeline, with scoring rules and a threshold.
  • Book a meeting — capture details, then offer your booking page.
  • Support triage — the issue, its urgency and an order reference, then hand to a person.
  • Start from scratch — an empty agenda.

A preset only fills in a starting point. Everything is editable afterwards, and nothing behaves differently because of which preset you chose.

How it behaves in a conversation

The rules are deliberately restrained, because a bot that interrogates people is a bot people close:

  • it answers the visitor's actual question first, then appends one follow-up — never a list;
  • it stays quiet for the opening turn or two (you set how many) so someone can explain why they came;
  • it never asks the same thing twice in a row, and gives up on a question after a couple of attempts;
  • it doesn't interrupt a conversation flow, and says nothing once a teammate has taken the conversation over.

It builds one record

As the bot learns things, it writes them to a single record — the first identifying detail matches or creates the contact, and everything after that is added to that same one. Two protections apply to every write:

  • It only fills empty fields. If a salesperson already typed a phone number, the bot leaves it alone. (You can mark an individual field as overwritable if you want the opposite.)
  • It can only write where you said. Each field names the CRM field it saves to, and that's checked against your real schema before the agenda is allowed to go live. Leave the box blank and the answer is collected for qualification but never saved.

Qualification

Score the answers, set a threshold, and optionally add conditions that must hold. Each conversation comes out qualified, disqualified, or unknown — and unknown is a real answer, not a polite no. A visitor who was never asked about budget is not someone who can't afford you, so an agenda set to wait for its required fields won't judge anyone early. Qualifying, disqualifying and completing each fire an automation trigger, so routing, notifications and follow-ups hang off them like any other event.

Publishing, and why there are versions

You edit a draft. Nothing you type affects a live conversation until you press Publish, which freezes the draft as the version every new conversation will follow. Conversations already underway keep the version they started with — so changing an agenda can never rewrite the rules mid-conversation.

Before publishing, Solstral shows you what's changing (fields added or removed, a moved threshold, changed actions). Publishing is blocked while there are problems — a field pointing at a CRM field that doesn't exist, a choice field with no options, or a score threshold with no scoring rules behind it, which nobody could ever meet.

Because every conversation records which version it ran under, you can always answer "what were the rules when this lead was qualified?" — which is why deleting an agenda that has actually run warns you first. Archive it instead.

Assigning it to a bot

On Chat → Bots, each bot has an Agendas row. Click to include one or more; they're pursued in order, one at a time. An unpublished agenda can't be selected — there's nothing live to pursue yet.