Aditya Khandelwal

Using a coding agent solo is one thing. Getting a whole team to trust agent-written code, agent-run reviews, and long-running agent work is another. That's where most teams stall. This talk is about what it actually takes to get there: how to shape a codebase so agents can work in it safely, how to earn a skeptical team's trust instead of mandating it, and the failure modes that only show up once agents are part of the daily workflow.