Write a useful brief
Treat the prompt like a teammate brief: audience, outcome, must-have fields, and tone. Vague prompts produce generic forms; specific ones produce editable drafts you can ship. Name the decision you need (qualify, diagnose, segment) and call out score fields such as NPS or CSAT when they matter - those map cleanly to Scale and related blocks in the catalog.
- Name the audience (trial users, post-purchase buyers, event attendees)
- State the decision you need (qualify, diagnose, segment)
- Call out score fields (NPS, CSAT) if you need them
Edit in the builder
AI generation is the starting point. Reorder fields, swap types across the available field types, add conditional logic with cycle detection, and theme the form. Browse each block in the form fields docs (/docs/blocks) and deepen follow-up behavior with AI follow-ups (/docs/features/ai-follow-ups). Logic & branching (/docs/features/logic-and-branching) covers show, skip, and jump rules before you publish.
Enable in-form follow-ups
Pick questions that benefit from clarification - low scores, open text, vague intent. Follow-ups fire while the respondent is still filling the form (public or authenticated). This is separate from Slack/Sheets alerts after submit. See the AI follow-up block (/docs/blocks/ai_followup) and the AI follow-ups feature doc (/docs/features/ai-follow-ups) for plan notes and how probes differ from after-submit routing.
Publish a version
Publishing creates a version. Submissions bind to the version they were collected on, so analytics stay coherent when you iterate. Read Forms, publish, and versions (/docs/features/forms-publish-versions) before sharing the public link or embed so draft edits do not surprise live respondents.