Series

Agent Architectures

Benchmarks, patterns, and deep dives into how AI agents are structured. AgentLab pits 4 architectures (ReAct, Plan-Execute, Reflection, Tree-of-Thought) against each other with real metrics. SwarmOS orchestrates 4 agents under a supervisor. RAG-Agent-Fusion grounds agents with verifiable retrieval. No demos — just architecture comparisons with data, trade-offs, and what breaks at scale.

Two LangGraph Projects That Taught Me How to Design Multi-Agent Systems
TL;DR — I built two LangGraph projects back-to-back: a Release Narrator with supervisor + hybrid interrupts, and a CI/CD Bottleneck Optimizer with a cyclic refine loop. 5 bugs shipped to production be
Jul 12, 202614 min read9

Command Palette