Blog

Thoughts on building reliable AI agent infrastructure.

2026-04-04 · 5 min read

Why AI Agents Need Durable Execution

AI agents that call LLMs, use tools, and make decisions over hours or days need more than a simple script. They need durable execution — workflows that survive crashes, handle retries, and resume from where they left off.

2026-04-04 · 7 min read

Building Reliable AI Workflows: Retries, Fallbacks, and Cost Control

LLM APIs fail. Rate limits hit. Models hallucinate. Here's how to build AI workflows that handle all of this gracefully — with automatic retries, provider failover, budget limits, and human-in-the-loop approvals.