Listen: Not all AI agents are created equal

Listen: Not all AI agents are created equal

From Lenny's Reads by Lenny Rachitsky

April 15, 2026 · 20 min

About this episode

Hamza Farooq and Jaya Rajwani discuss a framework for categorizing AI agent initiatives and overcoming common challenges in their development.

Hamza Farooq and Jaya Rajwani, who teach two of the most highly-rated courses on building AI agents, share a practical framework for categorizing and prioritizing your agent initiatives. Drawing on over 50 hours of research and work with Fortune 500 companies, including Jack in the Box, Tripadvisor, and The Home Depot, they explain why most teams get stuck when building agents—and how a three-category taxonomy transforms a confusing backlog into a clear, sequenced roadmap. --- In this episode, you'll learn: • Why prioritizing AI agent initiatives is so hard, and why familiar planning tools like impact-effort matrices break down • The three architectural categories every agent falls into • How to choose the right platform for each category • Success metrics and ROI frameworks tailored to each architectural type • How to quickly course-correct when you’ve picked the wrong architecture --- References • Agent Engineering Bootcamp: https://maven.com/hamza-farooq/agent-engineering-bootcamp • Agentic AI for PMs: https://maven.com/ • AutoGen: https://microsoft.github.io/autogen/ • Claude Code: https://claude.ai/code • CrewAI: https://crewai.com • Google ADK…

People in this episode

Host: Lenny Rachitsky

Guests: Hamza Farooq, Jaya Rajwani

Topics covered

  • AI agents
  • prioritization
  • frameworks
  • architecture
  • success metrics
  • course correction

Keywords

  • AI agents
  • prioritization
  • taxonomy
  • success metrics
  • architecture

Mentioned in this episode

Organizations: Jack in the Box, Tripadvisor, The Home Depot, Maven, Microsoft, Google, Gumloop, Lindy, Make.com

More episodes of Lenny's Reads

Explore listener stats, chart rankings, contacts and more on the Lenny's Reads podcast page.