AI campaign generation

Describe a goal in plain language and get a complete multi-step campaign — copy, timing, channels, and steps.

AI campaign generation lets you describe a goal in plain language and get back a complete multi-step campaign — including copy, timing, channel selection, and the steps that tie them together. It's the fastest way to go from an idea to a working draft.

Getting started

Open Campaigns → New, describe what you want to achieve under Describe your goal, and click Generate campaign — for example, welcoming new signups or winning back lapsed customers. The generator produces a full flow tailored to that goal.

Create a campaign page with the Describe your goal box
Describe your goal, then click Generate campaign.

Refine and accept

You don't have to take the first result as-is. You can iterate on it, refining the output until it matches what you have in mind. When you're happy, click Accept & Create Campaign to create a draft campaign that you can then edit freely in the flow builder — adjusting steps, timing, and the templates it uses.

Surveys and feedback

Ask for feedback in your goal, for example "Ask customers for an NPS score 3 days after delivery and tag detractors". The generator can then:

  • Make a new survey with up to 5 questions (NPS, CSAT, Yes / No, choice or text), or use a survey you already have.
  • Put a Survey block in emails, so the first question shows as one-click buttons.
  • Put a personal survey link in SMS, WhatsApp and push messages.
  • Add a Wait for event step that waits for the answer to that survey only.
  • Save the score to a contact field (for example last_nps) and branch on it: thank promoters, follow up with detractors.

Before you accept, the review shows each survey it will make, with its questions. New surveys are made and activated only when you accept. In-app surveys are not generated; add them yourself in an in-app takeover. See Surveys in campaigns for how each part works.

Generation history

Your generation history is saved, so you can revisit earlier attempts and compare different directions rather than losing them.

How it works

Generation is backed by a configurable LLM provider behind the scenes, so the underlying model can be adjusted for your workspace.

Next steps