Blog
Notes from the workshop. Things I built, broke, and learned from.
-
2026-07-24
The first feature was a pleasant place to begin
Early decisions from setting up a pleasant launch cockpit where a creative collaborator can run observable experiments without first becoming a project manager.
-
2026-07-23
The agent knew which plus button Sarah meant
A WhatsApp screenshot, a live CMS, a repository full of operating context, and four annotated images. This is what context engineering looks like when it helps a real person use new software.
-
2026-07-23
I do not use AI for one thing
Voice dictation, computer use, terminal agents, parallel work, client systems, prompt governance, private telemetry, and a way to check whether speed is improving or hiding my judgment.
-
2026-07-23
Dispatches and deep arcs
A 37-day trace of how I work with agents: hundreds of one-turn dispatches, a small number of deep systems arcs, and 6.4 billion cache-heavy tokens behind them.
-
2026-04-05
Building a Cost-Aware Intelligence Pipeline
How I built a system that tracks every penny it spends on AI - and why that's the whole point.
-
2026-04-05
AI-Augmented Photo Curation Pipeline
Building a resume-safe, multi-pass system that turns 4,700 unsorted iPhone photos into a categorised library - using metadata heuristics, perceptual hashing, and Claude's vision API.
-
2026-03-16
I invented a DSL for governance. Then I killed it.
Signal was a compressed notation for expressing process discipline. A 3-model adversarial test showed conventional shorthand works just as well with fewer characters. SD-321: Signal has no signal.
-
2026-03-01
33,700 tokens on boot and nobody noticed
I pointed a tokeniser at the repo. The boot sequence had an elephant in it.
-
2026-02-19
The agent that lied to protect me
I asked an agent to write about agents producing competent-sounding-but-wrong output. It produced competent-sounding-but-wrong output.