AI agents promise to automate everything from research and customer support to sales and business operations. But what happens when those agents start making decisions on their own?

In this episode of Today in Tech, Keith Shaw sits down with Postman co-founder and CEO Abhinav Asthana to explore one of the biggest unanswered questions in AI: can we actually control autonomous agents?

They discuss AI hallucinations, API security risks, MCP, agent-to-agent communication, accountability, emerging "agent manager" roles, and why the next generation of software may be harder to govern than anything we've built before.

Topics include:

* Why AI agents are gaining autonomy faster than companies can govern them

* The hidden risks of API keys, permissions, and machine identities

* How one agent's hallucination can become another agent's truth

* Why enterprises need guardrails before deploying agents at scale

* The rise of agent managers and AI governance teams

* Who is responsible when an AI agent makes a costly mistake?

If your organization is exploring agentic AI, this conversation highlights the opportunities, risks, and hard questions every business leader should be asking.