Setup
How recording and assistant setup work
Three ways to start, one editor at the end. Every path records a browser workflow and produces the same edited, narrated video that you review before export.
- 01
Describe it for the agent
Tell DemoGlow what to show and who the video is for. The agent plans the browser steps, records them and produces the first edit.
The agent needs access to the browser workflow you want recorded. Use a demo or staging environment you control. DemoGlow explains the supported access method and any manual step before recording starts.
Progress shows as:
- Planning
- Recording
- Producing
- Waiting for you
- Ready for review
- Couldn't complete
If something is missing, DemoGlow asks instead of guessing.
- 02
Record it yourself in Chrome
Install the DemoGlow Chrome extension and record the tab or window where the task happens. Speak while you work or stay silent; the script and narration are produced afterward.
Your recording is never altered. Cuts, speed changes and effects are decisions about it, and you can reverse each one.
Chrome on macOS and Windows. Desktop applications and existing video files are not supported.
- 03
Ask a connected assistant
Connect DemoGlow to a supported AI assistant through MCP. Request a video in plain words, follow its progress, and receive a link to the result.
The request uses your DemoGlow account and allowance. Review, changes and export happen in DemoGlow, not inside the chat.
Supported assistants are listed here once each one's complete path has been tested end to end.
Before you record
- Decide what the viewer should be able to do or understand at the end. One task per video works best.
- Prepare an account, data and environment that are safe to show. The agent records against the environment you authorize.
- Expect to review the result. The first edit is automatic; the decision to publish is yours.