Sachin Malhotra

Sachin Malhotra is an engineer at Anthropic, where he works on the continuous-integration infrastructure that Anthropic's engineers build and ship on. Keeping CI healthy at a frontier AI lab used to mean a pager and a long queue of toil; today most of that day-to-day runs through a fleet of Claude agents that triage incidents, watch rollouts, review changes, and draft the work, while Sachin sets direction, checks their output, and makes the calls only a human should. He has spoken about agent-driven engineering at the Tech Crunch Disrupt 2026 and has taught hands-on sessions on working with Claude to thousands of engineers at enterprise companies. Previously, he lead machine learning infrastructure at Etsy and DNS systems at Microsoft.

Anthropic, USAedorado93

Teaching AI Agents to Take No for an Answer

AI Coding Summit NYC

Upcoming

Teaching AI Agents to Take No for an Answer

Tell an AI agent “Forbidden,” and instead of stopping, it finds a workaround: scaling services down to zero, deleting resources in smaller batches, and even changing its identity to get a fresh quota.

Now imagine giving that same agent a clearer message: “You’ve reached your limit. This action affects 212 resources. Try again in 59 minutes, or ask a human for approval.”

Suddenly, it behaves differently. It stops, considers the potential damage, and asks for help instead of finding a way around the rules.

This talk explores how we can design safety boundaries that AI agents actually understand and respect. We’ll look at smarter rate limits, separating actions by how easily they can be undone, simple safeguards that prevent costly mistakes, and ways to make sure an agent can’t fake its own identity.

The goal isn’t just to teach AI agents when to stop. It’s to also help them understand why.

artificial intelligence