Nathan Crock

I like automating the process of inductive inference. To do that I study the systems that determine what an AI can know, decide, and do. My interests focus on three interlocking problems:

• Context Engineering — memory architectures, retrieval-augmentation, and context window optimization for foundation-model-based agents.
• Multi-Agent Orchestration — the coordination challenges that emerge when those agents must collaborate at scale.
• Decentralized Coordination — how autonomous agents reach agreement, divide labor, and exhibit coherent collective behavior through consensus protocols and peer-to-peer negotiation, without a central orchestrator.

These threads converge on a single question: how do we build a robust, composable, production-grade agentic web?