Research & analysis
Deep-dive research, real production case studies, and expert analysis on the failure modes that kill agent projects before they ship.
Latest research
Agents are overly optimistic planners. They commit to paths, underestimate subproblem complexity, and can't recover gracefully when the plan breaks. We studied 200 real agent failures and here's the pattern.
Context windows have limits, and most production agents hit them faster than you'd expect. Here's what actually happens when memory runs out — and what to do about it.
Our experts
AI Reliability Researcher
Sarah studies failure modes in deployed AI systems at scale. Before founding the How Agents Fail research collective, she spent six years at a major cloud provider building the infrastructure that runs millions of AI API calls per day — and watching them fail in increasingly creative ways.
Enterprise AI Implementation Lead
Marcus has shipped agent systems for Fortune 500 companies across three continents. He specializes in the messy middle: moving from prototype to production, where 90% of agent failures actually occur.
Our mission
We're building the definitive resource on how agents break — so the teams building them can ship faster, with more confidence, and fewer production disasters.
About the project