Skip to main content
Conversion AI is the AI layer built into Conversion. It can answer questions about your workspace data, build marketing assets, and diagnose issues across emails, forms, workflows, audiences, and campaigns. You can use Conversion AI three ways:
  • Chats: an interactive conversation you can open from any page in the product.
  • Agents: configured teammates that own a job and run on their own — triggered by Slack, Linear, a schedule, or a webhook.
  • MCP server: connect external AI tools like Claude, ChatGPT, and Cursor to your workspace.
All three run on the same underlying tools. Chats are how you answer a question once; agents are how you stop answering the same question by hand.

Capabilities

Limitations


How it works

When you ask Conversion AI to do something, it runs a series of tool calls: small actions like listing audiences, fetching an email, or creating a form. Each one appears in the chat as it runs.

Context

Conversion AI uses several sources of context when responding: Chats do not remember anything from previous chats. Each chat is its own conversation, with its own history of messages, tool calls, and edits. Agents add two more sources on top: their own instructions — standing guidance for the job they own — and the documents you upload to them, which they read as reference material.

More resources

Agents

Configured teammates that own a job and run on their own.

Glossary

Key terms used in Conversion AI.

Tool calls

The full list of actions Conversion AI can take.

Creating and editing

How Conversion AI creates assets and proposes edits to your assets.

Approvals

Which actions require a confirmation before they run.

Files and sandbox

What you can upload, and what Conversion AI can do with files.

MCP server

Connect external AI tools to your Conversion workspace.

FAQ

Common questions about Conversion AI.