​​

Core Frameworks

I've built conversational AI through four evolutionary stages. Each one taught me how humans and machines create meaning together.

​

The Four Acts of Conversational AI

Act I: Directed Dialogue — Rigid scripts. I learned guardrails.

Act II: NLU — Chaotic improv. I learned graceful failure.

Act III: LLMs — The writers' room explosion. Scale makes noise; design makes meaning.

Act IV: Agentic AI — The ensemble. Specialized agents, clear boundaries, choreographed performance. Responsibility designed in from day one. This is what I'm building now.

​

Stories Are Operating Systems

Before we had semiconductors, we had performance. AI is the next stage of humanity's oldest technology. Voice collapses the barrier between "user" and "programmer" because conversation is the programming language we all know.

​

The Human API

The language model is never the show—it's the noisy backstage. The real performance happens at the interaction layer. The human API is where meaning lives. It's not a dropdown menu—it's a constitution.

​

Rehearsal > Scale

Intelligence isn't in brute force. It's in rehearsal. The smartest systems aren't the biggest—they're the ones that fail gracefully and learn in the presence of an audience. Bigger models make noise. Better design makes meaning.

​

There's more to this story.

Bullet Holes & Algorithms: A Rebel's Guide to Voice AI
Coming 2026

​