AI engineering for production agents
What makes an agent reliable beyond a single successful demo: context, tools, evaluation, and feedback.
Agent optimization for better AI decisions
Better outcomes start with representative tasks, observable goals, and disciplined iteration.
Harness engineering for reliable AI agents
Repeatable runs, traces, and evaluation gates turn promising behavior into dependable software.
Context engineering for useful AI agents
Select the evidence, instructions, and permissions that belong in an agent's working context.
Notes
Field notes on software, intelligent systems, and the choices that make technology useful.