Free Lesson

Stop Flying Blind: Observability for AI Agent PMs

Hosted by Mahesh Yadav

Fri, May 29, 2026

4:00 PM UTC (45 minutes)

Virtual (Zoom)

Free to join

Invite your network

Go deeper with a course

Agentic AI Product Management Certification using Claude Code
Mahesh Yadav
View syllabus

What you'll learn

Understand Why AI Agents Fail Silently in Production

You'll learn why agents return 200 OK while failing — and which signals to watch before your users find out.

Map the 4 Pillars of Agent Observability

Understand traces, evals, monitoring, & guardrails — the four pillars every AI PM should own, not just hand off to eng.

Compare the Real Tools: LangSmith, Ai foundry , Braintrust

Instead of googling when your agent breaks, you'll know which tool fits your stack and what to set up before you ship.

Know Exactly What Is Your Job vs. Engineering's

Understand where your job starts: defining good, alert thresholds, trace triage. Stop handing it all to engineering.

Why this topic matters

Most AI PMs ship an agent and then wait for Slack messages to find out it broke. AI fails gracefully — confident wrong answers, no error codes, no alerts. Without observability, you're not managing a product, you're guessing. This session gives you the foundation to catch failures before users do and own quality as a PM skill, not just an eng task.

You'll learn from

Mahesh Yadav

Ex AI Product Lead -Google l Meta l Microsoft l AWS | 10k+ Alums l Founder - Agentic AI Institute

Mahesh Yadav brings 20+ years of experience building AI products at Google, Meta, AWS, Microsoft. He holds 12 patents in AI training, power management and computer vision. He has launched major agentic-AI initiatives (for example launching an agent for AWS Bedrock, featured in CEO keynote) and trained thousands of professionals to succeed in AI roles. With this programme you get rare access to CEO-level of mentorship.
Substack AI PM Newsletter l Linkedin Community of AI PMs l YouTube For Free Sessions Recordings
Currently, he is building back-office AI agents for the enterprise, starting with in-house legal teams through LegalGraph.AI. His work bridges education and real-world AI deployment, helping organizations adopt agentic systems that automate complex knowledge work.

Previously at

Google
Amazon Web Services
Meta
Microsoft
See all products from Mahesh

Sign up to join this lesson

By continuing, you agree to Maven's Terms and Privacy Policy.