Workshop Notes
Agentic software engineering shifts implementation to an AI system running its own loop. A working model of what changes: specification, orchestration, verification, and accountability, from 15 years leading engineering teams.
An AI product strategy decides which parts of a roadmap deserve AI investment and which are commodity features. A portfolio framework from 15 years leading product through eight acquisitions.
AI product discovery starts before the model: a real user pain, data that exists, the risk of a wrong answer, and workflow fit. A framework from 15 years leading product through eight acquisitions.
A robotics software engineer actually builds: perception, planning, control, and actuation. A maker's map of the stack and the software-to-hardware boundary, from 15 years in software and years on the bench.
Most technical due diligence checklists are lists of documents to request. Documents are the easy part. A seven-stage checklist organised around the questions that predict what owning this will actually be like.
Post-merger integration damages more good products than any other decision after close — usually through choices that were rational at the time. A frame for deciding what to consolidate and what to leave alone.
A personal operating system isn't a planner or an app. It's a four-step cadence, capture, orient, decide, review, built from 15 years of product work and a 36-hour mountain effort, for turning a pile of half-formed obligations into an honest next move.
The say-do gap is the distance between what customers tell you and what they actually do. A product story about being told no, shipping anyway, and watching the usage data disagree with everyone in the room — including me.
Technical due diligence usually fails on the questions asked, not the data gathered. A seven-stage frame — customer, product, technology, team, economics, risk, integration — from inside eight acquisitions.
Virtual reality simulations in healthcare education work when the goal needs presence, not whenever the budget allows a headset. A field guide to instructional fit, feedback, and transfer, from 15 years building the category.
Most AI rollouts skip to scale. A five-stage framework, readiness, a bounded pilot, workflow fit, evidence, scale, from 15 years leading product through acquisitions.
AI UX patterns aren't about chat bubbles or sparkle icons. Previews, reversibility, provenance, and clear escalation are what actually shorten the distance to trust, drawn from 15 years building clinical and platform products.
Decision-making under pressure isn't about staying calm. It's an orient-decide-act system you build before the pressure arrives, tested on product launches and on a mountain in Utah.
A POC proves a specific technical approach is possible before anyone commits to building it. What a proof of concept is, proof of concept vs prototype, how it differs from an MVP, and when to skip it.
Healthcare simulation technology isn't the hardware. It's rehearsal, feedback, facilitation, and confidence: the stack behind it, from 15 years building it.
This URL sold internet marketing services in Raleigh until the lights went out. Google never stopped asking for it. The honest answer, a decade late.
An MVP is the smallest version of a product that lets you test your core hypothesis with real users. Here's what it is, what it isn't, and when you shouldn't build one.
As of August 2026, AI hasn't replaced software engineers — but it has changed the work. An honest answer based on what I'm seeing in real teams, not vendor decks.
BEM is a CSS naming methodology from 2012 that outlived nearly every tool of its era. What Block, Element, Modifier still teaches in 2026, refreshed at its original address.
You don't hear the sorting happen. Part 1 of Five Feet of Headlamp: notes from 36 hours on a mountain in Utah, starting with the first honest act of the day.
Emotionally intelligent design isn't a feature. It's the way a website acknowledges the human on the other side. 15 years of building for the web, refreshed.
AI anxiety for developers is the fear of becoming obsolete as AI tools improve. It's not about the technology — it's about identity, craft, and value. An honest breakdown.
A return to publishing after a decade away. Why this notebook exists, what it covers, and how it gets written.
