AI agent observability is broken
Source: YouTube · Zack Korman · published Apr 9, 2026 · 17:22
The video critiques Octa's confusing "blueprint for the secure agentic enterprise" framework while highlighting that major AI providers like Cloud Code, OpenAI's CodeX, and GitHub Copilot offer audit logging capabilities for security monitoring 0:00-0:40.
Key Takeaways:
• Octa's framework for "secure agentic enterprise" appears disorganized with randomly placed AI agent terminology 0:00-0:11
• The framework includes elements like AI agent risk detection and human-in-the-loop controls 0:14-0:19
• The narrator finds the framework's structure unclear and questions whether the elements are properly related 0:19-0:23
• All major AI providers offer audit logging and telemetry for security monitoring except one unnamed provider 0:25-0:41
The video provides a critical look at enterprise AI security frameworks while examining the audit capabilities of popular AI development tools.
Sources:
Generate CPE Credits
Generate a professional CPE document from this video's transcript.
Estimated credit: 0.5 CPE hours
Estimate uses the video runtime (1 hour ≈ 1 CPE, rounded to the nearest 0.5, minimum 0.5, maximum 2.0). The final amount can be lower after review, never higher.
Topic: Cybersecurity. Commonly maps to: Security and Risk Management, Security Operations. Exact CISSP domains are assigned during generation.
Free account. One generation at a time, with a daily limit.
CPEBuddy is independent and not affiliated with or endorsed by ISC2, ISACA, or any certification body. Exports are formatted for common CPE submissions; acceptance is at your certification body's discretion.
Transcript Preview
First 800 characters of the transcript
This is Octa's blueprint for the secure agentic enterprise, which is like a really weird way of saying enterprise agent security. They decided to take every word they've ever heard about AI agents and just put them into random boxes. Where are my agents? Includes AI agent risk detection. What can they do? Includes human in the loop. I I don't understand this thing. Are these not supposed to be related? Like are these not columns? I don't know what I'm looking at. But as you can see, one of the things here is audit logs and telemetry. All of the big AI providers have audit logging capabilities. So you can see here that cloud code has monitoring. Open AAI's codeex has whole thing on observability and ways of getting data. GitHub copilot has audit logging capabilities. There is unfortunately …