AI-powered chatflows let you design complex conversational logic quickly, without starting from a blank canvas.
This guide walks you through the full process: from generating a chatflow with AI to managing live and paused flows.
1. Creating a Chatflow with AI Prompting
When creating a new chatflow, you can start in two ways:
Option A: Use a templated AI prompt
Click on a templated prompt (e.g. Password Reset Automator).
Select Generate to instantly create a structured chatflow with:
AI Mission blocks
Conditional success / escalation paths
API calls or assignments where relevant
This is the fastest way to get a production-ready flow that follows best practices.
Option B: Write your own AI prompt
Type your own instructions directly into the prompt field.
Describe:
The customer problem
What information the AI should collect
When to resolve automatically vs escalate to support
The AI will generate a complete chatflow based on your instructions.
2. Editing the Generated Chatflow
Once the chatflow is generated, it behaves like any other flow, you have full control.
You can:
Edit AI Mission instructions to refine tone or logic
Add or remove steps (API calls, assignments, close conversation blocks)
Adjust success, failure, or escalation paths
Rename the chatflow to match your internal conventions
Each block is editable, so the AI output is a starting point, not a limitation.
3. Rename, Save or Publish a Chatflow
At the top of the chatflow editor, you’ll find three key actions:
Rename
Change and rename the chatflow at any time
Save
Keeps the chatflow as a draft. Useful while iterating or reviewing internally.Publish
Makes the chatflow live and available to users.
This allows you to safely test, iterate, or stop automation without losing configuration.
4. Best Practices for AI Prompted Chatflows
Be explicit in prompts about when to escalate
Clearly define what counts as a successful resolution
Keep AI Missions focused on a single objective
Review related conversations before re-publishing changes