Raindrop launches CI-native agent simulation: test your AI agent before merge
soleio · x · 2026-09-18
Designer soleio shared the new landing page of Raindrop (raindrop.ai), a CI-native agent simulation tool: run your agent against realistic, stateful copies of your services before merging code.
The demo shows Raindrop running 20 scenarios against a pull request in 3m 42s, detecting 2 regressions and 1 warning, plus a cost-per-run delta ($0.042 → $0.051, +21.4%).
Key points:
- Bring your own traces: works with existing instrumentation like Langfuse and Helicone, no SDK rewrite
- Real code, unchanged: your agent's actual code runs against stateful service copies (e.g. a cloned Postgres orders table), letting you watch every tool call and state change without touching production
- The example shows an agent handling a refund dispute using Slack history, Stripe charge lookups, Supabase queries, and ClickHouse analytics
The product is currently in early access.
Related event: Raindrop Launches CI-Native Agent Simulation Testing(2 posts)→
More from coding & agent
- Give your coding agent a good feedback loop — output quality improves 10x — DanielLockyer · 2026-09-18
- Mollick: Claude Projects Spins Up 18 Specialist Agents to Crack Historical Mysteries — emollick · 2026-09-18
- Building a Real-Time AI Chat, Vision and Voice iOS App with Gemini 3.8 Live — amos_gyamfi · 2026-09-18
- A Correct Answer Can Still Invalidate Your AI Agent Eval, Microsoft Dev Blog Warns — WirelessLife · 2026-09-18
- Google and Speakeasy open-source their entire OpenAPI SDK generation suite — _philschmid · 2026-09-18
- Power User Hits Build Mode Usage Cap, Begs for Another Usage Reset — DBSteiger · 2026-09-18