Experiments
Operational experiments in workflow, tooling, and product systems.
A dated log of the tests I'm running, the constraints I'm hitting, and what changed my mind.
Agent-assisted delivery loops
What I'm testing
Testing where multi-agent workflows genuinely improve engineering output, and where orchestration overhead outweighs the gains.
Tools
Codex, Cursor, repo checklists, and parallel agent handoffs
Constraint
Useful speed only counts if review quality, integration discipline, and ownership stay intact.
What changed my mind
Parallel agents help most with repo orientation, first-draft implementation, and comparison work. The failure mode is orchestration theatre: too many agents, vague briefs, and more review debt than progress.
Next move: Keep the loop small: one clear owner, explicit checks, and tight handoffs instead of open-ended autonomy.
Related: The Velocity Paradox
Active
March 2026 to April 2026
Specification-first building
What I'm testing
Exploring how clearer intent, quality gates, and better prompts can turn AI tools into a forcing function for stronger architecture and communication.
Tools
Prompt scaffolds, acceptance criteria, and lightweight decision logs
Constraint
AI punishes vague intent immediately. Bad briefs turn into plausible but generic output.
What changed my mind
The biggest gain is not raw speed. It is forcing clearer thinking earlier, which improves architecture notes, pull requests, and conversations with humans as well as tools.
Next move: Turn the best specs into reusable templates for product and engineering work.
Related: What I am focused on now
Product-system sketches
What I'm testing
Working through smaller tools and interface ideas where product judgement, model behaviour, and backend reality need to line up cleanly.
Tools
Small Astro pages, interface notes, and API-shape experiments
Constraint
Interesting product ideas usually fail at the joins between interface, workflow, and backend reality.
What changed my mind
The strongest sketches are the ones that expose permissions, trust, handoffs, and maintenance load early. If those stay vague, the concept is not ready.
Next move: Publish more of these as concrete artefacts rather than leaving them as private notes.
Related: Selected work
What experiments are for
I use experimentation to test whether an idea survives contact with real constraints. That might mean trying a new AI workflow, prototyping an interface pattern, or pressure-testing a system boundary before it hardens into production code.
The goal is not novelty for its own sake. It is to build a sharper point of view about how software should be made, what quality still looks like in an automated environment, and where teams can get seduced by speed at the expense of trust.