A panel of enterprise security leaders including CISOs and AI startup founders discussed how to govern agentic AI without blocking developer productivity. Key themes included running agents in isolated sandboxes (Docker Sandboxes, MicroVM-based), keeping humans in the loop for dependency selection, using blessed images and minimum release age policies, and treating MCP servers as potential shadow IT requiring centralized governance. Docker's approach centers on portable, disposable sandboxes and an MCP Gateway as an enforcement chokepoint. The consensus: governance must be built in from day one, not bolted on after an incident.
Table of contents
Isolate, control, observeNanoClaw—a personal AI agentLaptops as the new prodPortability—from laptop to cloudWhen agents build the supply chainKeeping humans in the loopA training-wheels approachLayer security to limit the blast radiusMCP—the new shadow IT?Making it safe and easy to run MCP serversA once-in-a-generation opportunity—and challenge234 Impressions