10x Engineers Don’t Build 10x Companies | Sridhar Rajagopalan (VP, IBM)

Listen to full episode:

Summary: Open models are getting good enough to matter. Anant, Ed, and guest Sridhar Rajagopalan (VP, Platform Services at IBM) dig into what that shift actually means for engineering teams on the ground — infrastructure, security, and the practical stuff nobody puts in the pitch deck. Sridhar brings the view from the trenches: years of managing engineering teams have taught him that AI's biggest wins come from surfacing the tribal knowledge that lives in senior engineers' heads, clearing out process bottlenecks, and rethinking how teams are structured to work.

Chapters:

00:00 Opening and why open models matter now

00:31 Availability, reliability, cost, and security concerns

01:54 Microsoft, IBM, Nvidia, OpenAI, and Google back open models

03:36 The Open Source Security Alliance and open source security

06:16 Open models six months behind frontier — the case for self-hosting

11:28 Introducing Sridhar and team vs. solo AI productivity

16:31 "AI doesn't know what I know," the tribal knowledge problem

18:43 Conversational AI use over rigid prompt engineering

22:23 Why faster coding alone doesn't fix the org

24:42 Testing and review as the new bottleneck

30:41 Removing dependencies, approvals, and human-only handoffs

38:49 Automated task assignment and AI-driven team coordination

50:53 Documentation for AI, takeaways, and what's next

Sound Bites:

"AI doesn't know what I know."

"If you build a super highway into Manhattan, it's not going to make the traffic in Manhattan any better."

"The winners won't be companies with the fastest coders."