Unlock Agent Autonomy: The Runtime for AI-Native Systems — Tushar Jain, Docker
Agent safety, not intelligence, is now the primary blocker to deploying autonomous AI systems
“I don't think intelligence is the next big blocker for us to leverage agents. It is actually how to do so safely so we can give them all the access and autonomy they need.”
Docker's Tushar Jain argues the AI industry has largely solved agent intelligence and the harder unsolved problem is safe deployment — giving agents only the permissions they need to prevent unintended actions. He illustrates the stakes with a personal anecdote of an analysis agent that unexpectedly posted a private manager report as a public GitHub PR because it had write access it never should have had. The talk frames least-privilege access controls as a foundational primitive for production agentic systems.