Blog
Long-form notes on durable agentic systems, AI engineering, and building in production.
- Durable Statefulness in Agentic Systems: Keeping Agents on Track
How long-running AI agents stay on track across context rot, resets, and handoffs by separating state from memory and externalizing progress.
- Durable Autonomy in Agentic Systems: Catching Silent Failures
Your agent may finish successfully and still get the work wrong. Durable autonomy is how production systems decide when to proceed, pause, and ask for help.
- Durable Execution in Agentic Systems: Making Failures Irrelevant
Why production AI agents need durable execution: retries vs checkpointing, Temporal and LangGraph, and how to build agentic systems that survive crashes.
- Five Things I Noticed at AI Council 2026
Reflections from AI Council in San Francisco — on why agents keep breaking, demos vs. production, infrastructure over models, open source credibility, and the human-in-the-loop question.
- The Agentic Developer's Stack: How I 10x'd My Productivity Without Burning Out (or Burning Cash)
A practical first-person workflow using Spokenly, Opencode, and Zed to ship faster with agentic coding while keeping quality and API spend under control.