Matthew Dean
Writing
Essays on agentic delivery, AI engineering practice, and the review systems that keep the work accountable.
- From Personal Productivity to Trusted Delivery
- The enterprise AI knowledge base is becoming a shared knowledge plane
- Agents fail where human work has edges
- Skills are process infrastructure, not prompt snippets
- Local models are a routing decision
- Before switching models, inspect the context path
- Agent security starts with verbs
- The trace is the test case
- Your agent is only as good as its harness
- AI coding gains are won or lost at the review gate
- Stop hoping your AI assistant “gets it” Give it memory and a structured process