Engineering disciplines for the AI-native era.
The methodologies, frameworks, and infrastructure patterns that define how high-performing teams build with AI agents.
Index
Three disciplines, one stack
Each concept maps to a layer of how LoomStack thinks about AI-native engineering. Start anywhere — they connect.
Agentic Engineering
Building systems where AI agents operate as autonomous, coordinated actors. Multi-agent coordination, failure modes, and the infrastructure that makes agents reliable at scale.
Harness Engineering
Evaluation and governance infrastructure that constrains agents toward verified, policy-compliant output. Eval harnesses, governance harnesses, and the two-layer framework.
Loop Engineering
Designing automated systems that prompt your agents instead of you prompting them directly. The shift from prompt engineering to systems that run, verify, and iterate on their own.
How they connect
Coordinate
Route agents, humans, and systems with shared context and policy.
How they connect
Constrain
Gate output with evals and governance before it reaches production.
How they connect
Automate
Design loops that prompt agents so you do not have to.