Skip to main content

Getting Started with AI task follow-ups (aka Agent Chat)

Ask questions, request changes, and dive deeper into your task outputs with interactive AI-powered follow-ups.

Joel Charboneau avatar
Written by Joel Charboneau
Updated today

Overview

After any AI task runs, you can ask follow-up questions to clarify results, request changes, or explore next steps. It's like having a conversation with the AI about your output. We call it Agent Chat.

How to use Agent Chat

Run a new task or go to Inbox β†’ Click an AI task output β†’ Start typing in the chat box

The AI will respond to your question and show you the steps it's taking along the way.

Example questions

  • "How does this compare to my brand?"

  • "What should I do next?"

  • "Can you create a Slack-ready synopsis?"

  • "Turn this into a creative brief."


What is Agent Chat?

Agent Chat is for serious analysis and building on your ideas. Think of it as "thinking mode" and a space to expand on concepts or take the next steps from idea --> execution.
​
Agent chat includes:

  • In-depth responses stringing together multiple tasks (as needed)

  • Complex queries (sometimes taking over 10 minutes)

  • Comprehensive, high-quality analysis

  • Production-ready outputs like brief outlines


Adding context for better results

Upload files to guide the response:

  • Example ads, brief templates, or brand guidelines

  • Screenshots of LinkedIn posts or strategies

  • Any relevant materials

This extra context gives you more control over the output and helps deliver exactly what you're looking for.


Understanding the Agent Chat response process

Agent Chat shows you its steps in real-time as it:

  • Runs multiple tasks

  • Compares expert outputs

  • Reference templates and content from Motion

  • Analyze and synthesize information

Tips for best results

  • Be specific with your questions

  • Add relevant context files to guide the response

  • Continue the conversation to refine outputs

  • Provide feedback on responses to improve the system

Did this answer your question?