Securing AI Agents in Production | REIN Security at TechCrunch Disrupt | What Breaks Next

Matan Bar-Efrat Rein Security CEO: AI Agent Security AI agents are moving from pilots into production. They call APIs, run code, touch databases and move money. But most security tools still watch the network or the container, not the agent itself. In this episode of What Breaks Next, Megan Dahlgren of Mesa Capital Partners sits down with Matan Bar-Efrat Rein Security CEO, at TechCrunch Disrupt 2026 to talk about what it really takes to secure agents in production. They cover: why Rein runs as a sidecar to the agent why one org-wide policy ends up either too lenient or too harsh how self-hosted deployment keeps sensitive data in your environment the agentic attacks Rein is already seeing in the wild They also get into how Rein goes beyond alerts: it detects abnormal behavior, prevents it with a learned allow list, and proposes a pull request to fix the root cause. A conversation for CISOs, CTOs, and anyone building or securing AI agents. Chapters 0:00 What RAIN does: securing agents in production 1:20 Granular policy vs. one-size-fits-all 2:29 What a deployment looks like 3:19 Setup, AI sovereignty, and time to value 4:18 Daily CISO reports, developers, and forensics 5:25 Agentic attacks RAIN is seeing 6:10 The real state of AI agent security 6:52 Detect, prevent, remediate 7:46 One message for CISOs Links 📖 Read the write-up: https://mesapartners.io/blog/rein-sec... 🛡️ Learn more about REIN: https://reinsec.io/ 🤝 Work with Mesa Capital Partners: https://mesapartners.io 🔗 Connect with Megan on LinkedIn:   / dahlgrenmegan   👉 Subscribe for more What Breaks Next conversations on what enterprises will actually adopt in AI infrastructure and security. 💬 Running agents in production? Tell us what’s keeping your security team up at night in the comments. #AIAgents #AgenticAI #Cybersecurity #CISO #AISecurity #AIInfrastructure #WhatBreaksNext #TechCrunchDisrupt26