Notes from building agent systems

Evaluation harnesses, orchestration patterns, memory design and the failures behind them — written up while the details are still fresh.

All posts llm-agents 4 agent-safety 2 evaluation 2 multi-agent-systems 2 orchestration 2 agent-economics 1 agent-evaluation 1 agent-failure-modes 1 agent-governance 1 agent-memory 1 agent-sandboxing 1 agent-security 1

Automated assembly line with robotic arms positioned along a conveyor in a factory

20 Sept 2026·6 min read

The Chief-of-Staff Pattern: When to Orchestrate Agents and When to Just Write a Script

Multi-agent orchestration pays off only when subtasks need different context, different tools, or independent verification. Otherwise it is a slower function call.

multi-agent-systemsorchestrationllm-agents

Close-up of a 3D printer nozzle laying down fine layers of filament on a build plate

20 Sept 2026·5 min read

Memory Is a Write-Policy Problem: Retention, Provenance and the Cost of Remembering

Most agent memory failures originate at write time, not retrieval. Provenance, sensitivity classes and schema-level expiry fix them earlier and cheaper.

agent-memorywrite-policyprovenance