Eps.75 | Technology: The bouncer for your AI agents: Barndoor.ai's Oren Michels on securing digital workers

"Agents are like enthusiastic employees who don't have a conscience and aren't afraid of getting fired." This stark warning from Oren Michels, CEO of Barndoor.ai, captures why enterprises desperately need governance for their AI workforce. With documented cases of ungoverned agents accidentally deleting entire code bases, the stakes couldn't be higher.Drawing from his experience founding Mashery (acquired by Intel), Oren recognised that AI agents face the same governance challenges as APIs did 15 years ago – but with far greater risks. He reveals why Anthropic's MCP protocol lacks built-in security, explains the four critical vectors of agent security, and shares why Moderna's HR department is leading their AI agent strategy. For any organisation deploying AI agents, Oren's insights on treating them as digital employees rather than traditional software could mean the difference between transformation and catastrophe.Guest bioOren Michels is CEO and co-founder of Barndoor.ai, pioneering security and governance for AI agents. Previously, he founded Mashery, an API management company acquired by Intel, where he served as GM of Intel's API Management Group. His experience scaling API governance to hundreds of enterprise customers now informs his approach to securing the next generation of digital workers.Topics covered• Understanding MCP and why agents need different protocols than traditional software• The four vectors of agent security: agent identity, human identity, tool access, and tasks• Why agents are "enthusiastic employees without a conscience"• HR and IT collaboration to manage agents as digital employees• Building platforms showing which humans achieve value from agents• Documented failures: ungoverned agents deleting entire code bases• Moderna's HR-led approach to AI agent deployment• Evolution from internal use to cross-company agentic workflows• Future implications for early career professionalsKey timestamps00:00:00 – Introduction to securing AI agents00:02:13 – Why API management patterns apply to AI agents 00:07:15 – MCP protocol and its security gaps00:12:14 – What Barndoor does00:20:06 – Documented failures and ungoverned agents00:23:39 – Moderna's HR-led approach00:26:09 – HRIS for the agent workforce00:34:43 – Future of work and early career challengesResources mentioned• Oren Michels on LinkedIn – CEO of Barndoor.ai • Tim Bond on LinkedIn – Podcast host • Barndoor.ai – AI agent security platform • Intel – Acquired MasheryListen and subscribeFound this valuable? Subscribe on: Spotify | Apple PodcastsJoin our LinkedIn community for ongoing discussionsQuestions? Email: [email protected] Learn more: https://techpros.io/Full transcript available below.

There are no comments yet.

Please log in to write the first comment.