How we verify use cases
Most lists of “AI agent use cases” are ideas, not deployments. We only list things that happened, and we show you the proof.
Every entry has
- At least one public source link, checked by a script on every update.
- The reported outcome, in the source's own numbers.
- An evidence label that says who made the claim.
- At least one caveat: what the numbers do not tell you.
Evidence labels
- Vendor case study
- Published by the company that sells the agent. Treat numbers as marketing until confirmed elsewhere.
- Company reported
- Stated by the company that uses the agent (blog, earnings call, interview). Not independently audited.
- Press report
- Reported by a news outlet, usually based on company statements.
- Research paper
- Described in a paper with a stated method. Peer review status varies.
- Independent
- Checked or measured by a party with no stake in the result.
Starter prompts
Prompts on use-case pages are written by us as a starting point. They are not the prompts the organization used, unless the page says so.
Sponsorship
Sponsored placements are labelled “Sponsored”. Sponsors cannot change the content, evidence label, or caveats of any entry.
Corrections
Found an error? Send a correction with a source. We fix mistakes fast and note the update date.