Why Most Agentic AI Systems Fail in Production — A Software Architect's Perspective
Old engineering principles aren't optional. They're load-bearing
May 28, 202611 min read123
Command Palette
Search for a command to run...
Series
Architecting Agents
Agentic AI is the most exciting shift in software in a decade — and one of the easiest to get wrong in production. Architecting Agents is a series of deep-dive essays from a senior software architect's chair: why agentic systems fail under real-world load, which classical distributed-systems patterns still apply, and how to design agents that survive contact with production. Less hype. More engineering. Written for engineers who ship.
Architecting Agent Memory: A Distributed Systems Perspective
The hardest problem in agentic AI isn't intelligence. It's memory.
Jun 3, 202612 min read24
Reasoning Is Cheap, Side Effects Are Forever
Why safe agentic AI is a distributed systems problem we already solved
Jun 4, 202615 min read7
In Probabilistic Systems, You Watch the Shape of Success
All green. All wrong..
Jun 9, 202618 min read35
Your Agent Has a Spending Problem
Cost is an architectural property, not a billing surprise. Treat your agent as a metered process, or a few sessions will eat a month of budget in a week
Jun 11, 202615 min read10