With the rise of AI, more code is being written than at any point in history. That's a triumph for shipping and a crisis for fixing. The truth underneath the agentic boom is that the same systems generating code at superhuman speed are structurally blind to what that code does once it's running. An agent can write a thousand lines it will never see fail. When no human wrote the code either, the oldest debugging instinct, “I think I know where this broke,” is gone.
In this fireside chat, Sentry CEO Milin Desai makes the case that having production context is the key to solving this crisis. Things like errors, traces, real user impact and the signal agents are not able to generate for themselves. Consequentially, they are the only thing that tells them what to actually fix. Drawing on Sentry's work as a launch partner for Anthropic's Claude Managed Agents and on Cursor's event-triggered Automations, Milin will explain how to close the loop in real time. He’ll explain why infrastructure builders like Stripe are wiring this context layer into the AI-native stack.
After defining the solution, he’ll address what happens when the gap between "something broke" and "here's the fix" collapses from hours to seconds. He’ll explore how when debugging starts to disappear and self-healing becomes more prevalent, should the loop fully close or do we still need a human in it? Why?