Turn screen recordings into repeatable procedures

They used Screenpipe to log screen and audio locally, then had an agent convert repeated work into SOPs and automations.

Done with Custom agent· Intermediate· Jul 2026

What they did
They ran Screenpipe in the background so it could notice meaningful changes like app switches, clicks, and typing pauses. When that happened, it saved a screenshot and the accessibility tree, used OCR if needed, and transcribed audio on the device. All of that was indexed locally, and an agent could then search it through the API, MCP, and skills to spot repetitive work and draft procedures or automations.
How it went
The post does not report any finished automation results or performance numbers from a real deployment. It does say the local models are intended to stay under 1% CPU use and 400 MB of RAM.
Tips
This works best if you let the recorder run long enough to capture a real routine. If you plan to automate anything risky, have the agent stop and ask before any action that sends, submits, pays, or deletes.

Try it yourself with Claude Code

I want to turn my repeated work into a procedure and then automate the safe parts. My task is [describe task] and I use [apps/tools]. Review the activity you can find, identify the repeated steps, and write a clear step-by-step SOP in [folder path]. Then create a script or reusable instruction file that handles the repetitive parts. Test it on a harmless example first, and stop for my approval before anything that sends, submits, pays, or deletes.

Fill in the [brackets]. Agents can make mistakes, so review anything before it is sent, paid or deleted.

Get 3 agent uses like this every morning

With the day's top AI news. Free, and one click to unsubscribe.