Field Notes
Field reports from inside live systems
Each note starts with a specific incident, the decision made, and the edge of what the evidence can support. If a source cannot carry a conclusion, the conclusion stays out.
Every note keeps first-hand context, a limitation, and a route back to the relevant problem or case.
Four operating domains
I build each hub around one operating problem. It is not an automatic tag page.
delivery
Delivery field notes
Reports about restoring a release path, decision ownership, and a handoff the team can keep.
Open the hub: Delivery field notesai-engineering
AI engineering field notes
Reports about putting AI into engineering work without losing review, provenance, or technical ownership.
Open the hub: AI engineering field notesproduct-risk
AI product risk field notes
Reports about product boundaries, ambiguous signals, safety controls, and evidence that survives scrutiny.
Open the hub: AI product risk field notesresilience-security
Resilience and security field notes
Reports about systems that fail under real constraints and the decisions that make recovery possible.
Open the hub: Resilience and security field notes
Notes to start with
A short editor-selected set of published work. Each domain hub keeps the complete same-language list.
Resilience & Security
The throughput target no customer needed
A field report about sizing a system for a demand assumption that the business never forced itself to validate.
Read the note: The throughput target no customer neededAI Engineering Control
Two roadmaps make fresh code feel like legacy
What a repository reveals when the team's real planning horizon is shorter than the official product plan.
Read the note: Two roadmaps make fresh code feel like legacyDelivery Recovery
An AI transformation needs an owner
Acceleration language means little until one person owns the operating boundary and its trade-offs.
Read the note: An AI transformation needs an ownerAI Engineering Control
Agents work better as advisors
A practical boundary for agentic engineering when confident automation is still wrong often enough to matter.
Read the note: Agents work better as advisorsAI Product Risk & Safety
AI product risk lives in the gray zone
Why a plausible answer is not enough when the product must detect ambiguity and stop the normal flow.
Read the note: AI product risk lives in the gray zoneDelivery Recovery
A fractional CTO must not leave an empty chair
A fractional mandate works when it transfers ownership instead of renting one person's judgment.
Read the note: A fractional CTO must not leave an empty chair
A feed without an algorithm
RSS carries the published English notes.