Anyone can build an agent. Who can build a reliable one?
Sri Kandikonda runs FedEx’s API ecosystem across 200+ countries, and his product team works alongside AI agents every day. He explains why most agents die at proof of concept, and what the ones that ship have in common.
Sri Kandikonda
Group Product Manager, FedEx
15+ years across e-commerce, logistics and fintech. Teaches a Maven course on taking AI agents to production and wrote a book on the API economy.
Follow Sri on LinkedIn →
Anirudh Narayan
Co-founder & Chief Growth Officer, Lyzr
Co-founder & Chief Growth Officer, Lyzr. Host of How We Agent. He built the agent that pitched Lyzr to its Series A investors.
Follow Ani on LinkedIn →Must watch if your agents keep stalling at proof of concept while your AI bill keeps climbing.
Chapters
Orange dot marks a key momentFive moments worth pressing play for
Short on time? Pick up where you want.
-
Moment 01
One internal agent answers 80% of product questions
How Sri’s team built it, and where it hands off to a human instead of acting on its own.
-
Moment 02
Token maxxing
The hidden cost of AI agents, and why AI bills can climb from $50K to $15M a month.
-
Moment 03
Write the agent spec first
Start with the business case, and don’t ask one agent to do everything.
-
Moment 04
Four things every production agent needs
Miss one and the agent stalls at proof of concept:
- A workflow owner
- Evals
- Data governance
- Cost governance
-
Moment 05
Internal before external
Why FedEx builds agents for its own teams first, and why every agent needs reliable, real-time data and APIs.
Information about the shipment is as important as the shipment.
Up next
All episodes
Up next

A Breakdown of What Cisco’s AI Infra Stack Looks Like
Raj Rikhy, Senior Director of Product Management
Want more of this?
Get notified when the next episode drops, read case studies or get AI agent use cases and templates right in your inbox every month.
No spam. Unsubscribe anytime.

