Blog
Introducing Raindrop 2.0: Self-Healing Agents
When your agent fails in production, Raindrop detects it, your coding agent fixes it, and the failure becomes an eval so it never happens again.
Recent Posts
How Speak uses Raindrop to build better agents for 15 million users
Speak relies on Raindrop to surface the long tail of issues in their AI language tutor, protect the high-trust learning experience, and ship improvements for over 15 million users across 40+ countries.
Introducing Raindrop Workshop
The first sane way to debug your AI agent locally. Open source, MCP-native, one command to install.
How GC.AI Closes the Eval Loop with Raindrop Workshop
GC.AI uses Raindrop Workshop to inspect traces, validate eval pipelines, and let their coding agent run the entire benchmark loop from a single session.
More Articles
Introducing Raindrop Triage
An agent that investigates your AI agents. It lives in Slack and Web, it's also an MCP, and it's the foundation of what comes next at Raindrop.
Introducing Trajectories
Visualize, search, and debug agent traces. Trajectories is a purpose-built trace viewer for AI agents - with natural language search, duration and output-size visualizations, and AI-powered trace explanations.
Introducing Agent Self Diagnostics
Your AI agent already knows what's going wrong. Now it can tell you. Self Diagnostics lets agents proactively report failures, loops, and capability gaps giving your team visibility into problems that would otherwise go silently undetected.
Announcing the Query SDK
An agent-native toolkit for AI observability data that goes far beyond traditional query languages, with semantic search, signals, timeseries, and more.
Announcing Our $15M Seed Round
We're excited to announce $15M in funding from Lightspeed and leading AI companies to build the monitoring platform for AI Agents.
How Spiral solved a critical launch bug with Raindrop
Spiral used Raindrop to rapidly resolve critical tool errors during their V3 launch, eliminating an issue affecting 1/3 of users.
Real-Time Fixes, Real-World Trust: How Howie Uses Raindrop to Eliminate User Frustration
Howie's team prioritizes customer input by reviewing every Raindrop User Frustration alert and implementing immediate fixes, reducing churn by 28%.
Thoughts on Evals
The real test for how good your product is how it performs in the real world.
How Tolan's AI Engineers Monitor Their 'Alien' Companions with Raindrop
Tolan reduces memory issues and lore inconsistencies by monitoring issues and validating fixes.
How Unstuck AI Monitors Their Learning Agents with Raindrop
Unstuck prioritizes fixes with daily alerts and validates improvements with incidence monitoring.
New Computer Monitors Dot's Memory and Personality
How New Computer keeps Dot reliable by surfacing issues early and validating fixes with Raindrop.
Raindrop Achieves SOC 2 Type II and Introduces Raindrop Notify
Announcing SOC 2 Type II certification and Raindrop Notify for privacy-first observability.
o1 isn’t a chat model (and that’s the point)
How Ben Hylak turned from ol pro skeptic to fan by overcoming his skill issue.