Observed arrival · 2026-10-12
Rehearsal puts AI agents through practice runs on your app
The tool builds a sandboxed copy of an application with fictional data, runs AI agents through jobs, and checks the outcomes against the app’s database.
○Open
⊠Login
$Paid
†Ads
✦Pretty
●Pro
◎Niche
◉Human
⚑Risk
ƒJS
- For
- Teams evaluating agents that act on application data
- Worth noticing
- The displayed Gitea support-agent example uses 48 held-out episodes and shows 71% verified success against a stated 90% pilot threshold.
Field notes
The described build sequence runs from reading source and starting a sandbox to seeding fictional data, selecting tools, and publishing scenarios. Rehearsal says it uses read-only checks against application records rather than an agent’s own account of success. Its example readiness panel reports results across 48 held-out episodes, while the page marks custom-model training as a future capability.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
○OpenPublic substance visible
$PaidCommerce or pricing visible
✦PrettyNotable craft visible
●ProPolished or operationally mature
◎NicheUnusually specific use
ƒJavaScriptBrowser-side code central
One card from the complete issue