My learnings from the Max Agency podcast with Izzy Miller—how Hex evolved from single-shot SQL to a full multi-agent notebook platform, and what it took to get there.

An exploration of long-horizon AI evaluations like Vending-Bench 2, demonstrating where modern LLMs thrive and break down over year-long operations.

Exploring Pi-mono, a minimalistic coding agent, and pi-mom, a self-managing Slack bot that writes its own tools.

From 'vibe coding' to structured spec engineering and harness engineering as the foundation for building with AI agents.

Shadow AI agents - can the business function if a given agent stopped working? The new automation debt.

A deep dive into the 'Head of AI Observability' persona and a product requirements document for an agent governance platform.

Exploring the impact of AI coding tools on developer practices and the risk of accumulating AI legacy code.

Exploring the challenges of monitoring AI agents at scale and the need for a unified observability and governance platform.

Anthropic's study found that AI assistance hinders skill development for developers learning new technologies.

A behind-the-scenes look at how this blog was architected and built.