The right details. In the same conversation.
A quote request, a callback or a support case needs a few clear details. Let the assistant offer a relevant published form for visitors to complete inside your website chat. Your team handles the answers with the conversation in view.
Forms are available on every plan, with a limit set by your plan. Manual sending uses Take Control, available from Pro and during an active trial.
Website chat
I'd like a quote for a team workshop.
You can send us the details here.
Workshop enquiry
- Name
- Alex
- alex@example.invalid
- Your request
- Could you send a quote for a workshop for our team?
Submitted values stay out of the AI assistant’s model context.
Your fields. Your languages. Your request.
Build the form in Content, finish its confirmation and privacy details in Settings, and describe when to offer it in Offering.
Start from a useful template
Choose contact, quote request, appointment request, support, complaint, callback or feedback. Or build your own form from scratch.
Arrange fields by dragging
Move fields, pair two side by side, and edit labels and required status. Where language versions are available, prepare their text on the same field structure. Check the regular and narrow widget previews.
Publish and choose when to offer it
Write a short offering instruction, such as a request for a quote or a callback, then publish the reviewed version. The assistant can choose it during a relevant conversation. An operator can also send it through Take Control.
Workshop enquiry
An example of details a visitor has already submitted to your team. All details are fictional.
- Name
- Alex
- alex@example.invalid
- Your request
- Could you send a quote for a workshop for our team?
Submitted values stay out of the AI assistant’s model context.
Responses your team can work through.
Submitted details belong in a clear overview, not scattered messages. Read the answers, open the related conversation and keep track of what has been handled.
| Submitted | Request | Status | |
|---|---|---|---|
| Today, 10:24 | alex@example.invalid | Workshop quote | Pending |
| Yesterday, 16:40 | sam@example.invalid | Callback request | Processed |
- Keep the conversation close
- Open a response from its form card in the conversation, or move from the responses table back to the chat.
- Separate pending from processed
- Mark responses as processed individually or in bulk. Use the filters to focus on pending work, and return a response to pending when needed.
- Send answers to your systems
- Connect a webhook to forward submitted answers to your CRM or a Make or Zapier workflow. Select Form submitted (form.submitted) for the assistant. Webhooks are available from Standard and during an active trial.Set up form webhooks
- Keep submitted values out of model context
- The assistant receives the form's title and status, not the filled-in values. Authorized team members can read the responses and erase their values when needed.
Before you send your first form
A few useful boundaries for setting up your team's workflow.
- How does a form reach the visitor?
- In Offering, describe when to use the form and publish it. The assistant can then select it during a relevant website conversation. An operator holding Take Control can also send a published form from the composer. Your plan sets the number of forms. Manual sending requires Take Control on Pro or higher, or an active trial.
- Can one form have several languages?
- Yes. The form shares its fields, layout and validation across the assistant's configured languages. Write the labels, choices and confirmation text for each version, then confirm your review before publishing. The form uses an available published version for the widget content language, with the default language as a fallback. Text is not translated automatically.
- Does an appointment form make a booking?
- It collects a preferred date and contact details for your team to review. A submission does not reserve a slot or run a Custom Action. With an explicitly configured form.submitted webhook, you can forward the answers to your system, which handles any further action.
Ask for what your team needs.
Prepare a form for one real customer request, check its widget preview and decide how your team will handle the responses.