A record of what agents actually do
Most talk about AI agents is about what they might do. This directory collects what people have already done with them: specific tasks, the agent that handled them, and the prompt that made it work.
Each entry links to where it came from, so you can judge it yourself. We don't rank agents against each other or accept payment to list a use case. Sponsors are labelled and kept out of the listing.
Right now there are 21 use cases across 7 categories and 9 agents. New ones are added every week.
What counts as a use case
A real task, completed by an AI agent, described well enough that someone else can repeat it. Demos, benchmarks and product announcements don't count on their own.
How an entry gets published
- Step 1
Submitted or found
Entries come from the submission form and from public posts, threads and write-ups.
- Step 2
Source checked
We confirm the source exists and that the task was actually carried out, not just proposed.
- Step 3
Prompt cleaned up
Personal details become [placeholders] and the steps are written so someone else can follow them.
- Step 4
Published
The entry goes live with the agent, category, source link and the date it was added.