Greg Whalen

In this episode, Keith sits down with an AI infrastructure leader to unpack what is really happening behind the current AI boom.

While most conversations focus on models and hype, this discussion goes deeper into the engineering realities of building reliable AI systems at scale.

The conversation explores why observability, telemetry, debugging, and system architecture are becoming critical as enterprises move from experimentation to production AI.

They also discuss why terms like “agentic AI” are often misunderstood, why many AI projects end up being rebuilt from scratch, and why the future competitive advantage may come from the tooling built around models rather than the models themselves.

Topics covered include:

• Why most AI systems fail due to poor observability
• The real engineering challenges behind enterprise AI
• Why “agentic AI” may be the industry’s most overused term
• The shift toward commoditized foundation models
• Why telemetry and debugging will define reliable AI systems
• The underrated founder skill of ignoring distractions
• What the AI stack could look like by 2028

This episode is a practical conversation about the difference between AI demos and real AI infrastructure.