Blog

Notes from the work

Decisions, trade-offs, and what held up in production. Filter by category, or read everything.

Product Building

When to refactor instead of patching: what a bug that keeps coming back is telling you

A bug you keep fixing in the same place is usually a design problem. How to tell when the patches point at the architecture, through two rebuilds: a Gravity Forms integration moved to one state store, and an Asana and GitHub automation turned into a state machine.

Design × Code

How to encode an idea in a design: the concept inside my home page's particle field

The particle field on my home page is marked as decoration. Read rule by rule, it encodes one idea about how people connect, and a method for building a concept into any design.

AI Leverage

Single source of truth in project management: one brain your team and its AI agents share

A team with three places to check has none. How to run project work from one shared brain, a tracker extended with the repository's docs, that people and AI agents both feed and read, tied to the roadmap, firm hour to hour and flexible day to day.

Product Building

What a SOC 2 Type 2 audit asks of a small engineering team

A Type 2 audit checks that your controls ran every day for months. What that asks of a small engineering team, and which parts cost the most.

Product Building

Audit proof your deploy pipeline leaves behind

The cheapest SOC 2 evidence is the kind nobody gathers by hand. How a deploy pipeline can leave it behind as a side effect of shipping.

Product Building

Build or buy automation: when to build in-house and when to subscribe to a SaaS

AI made building internal automation cheap. It did not make owning it cheap. The criteria for deciding between building in-house and subscribing to a SaaS, for dev and ops tooling and for AI agent workflows, with the labor arithmetic most teams skip.

Product Building

SOC 2 Type 1 versus Type 2: what the auditor actually checks

Type 1 checks that your controls are designed. Type 2 checks that they ran, every day, across a window of months. Which one buyers ask for, and what each asks of engineering.

Product Building

The role of DevOps: protect developer and AI agent capacity

DevOps exists to protect the team's capacity: the developers' hours and the AI agents' runs. How matching environments keeps both spent on new work, and why the honest zero is no bug an earlier environment could have caught.

AI Leverage

Untangle: how to plan a few weeks of team work with Claude

Put the work in a dependency graph, have Claude walk it twice, once as if nothing is done and once as if every goal was reached, then bundle the tasks so people hand off less and learn more.

AI Leverage

A Claude Code setup that improves itself: scripts for the fixed steps, skills for judgment

Keep the global setup in git, move every repeating step into a small parametrized script, and leave markdown only the judgment. How a goal-driven setup turns each failure into a rule, and each rule into code where it can.

Product Building

How to weigh competing priorities: put them on a circle

Quality, cost and speed pull against each other, and a ranked list hides how. Plot each priority on a circle, place them by how much they agree, weight their pull, and see where the work lands and what it moved away from.

Product Building

Local, staging and production: what has to match, and what staging is for

Three environments, three jobs. How to set up local, staging and production, what has to match between them and what must differ, and why a staging run of a critical change should teach you something rather than prove you right.

Product Building

AI product strategy: every team got wider, few went further

AI made it cheap for a small team to offer more, and most are spending that on ground someone already covers. Why widening looks like Google's graveyard, and why cheaper execution is a reason to go deeper.

AI Leverage

AI code quality: clever code got cheap, clever decisions did not

Maintainable code was insurance against the cost of changing it later. AI collapsed that cost, so the backend can carry plain, improvable code, and the cleverness belongs where the customer meets the product.

AI Leverage

Developer productivity metrics after AI: what they can't see

AI made engineering output cheap to produce and easy to count. What productivity metrics miss about honesty, hidden effort and motive, and how to make each one leave a trace.

General

Ready is a finish line. Better is a direction.

Looking at work and seeing how it could improve feels like the same act as seeing what it takes to ship. It is the opposite act. Count the steps, box the effort, and cross the line.

AI Leverage

SaaS UX with AI: stop presenting options, guide to the outcome

Menus were a workaround for software that could not understand a sentence. That constraint is gone, and most SaaS interfaces have not noticed. What changes when the product knows the user's next few tasks and walks them through.

AI Leverage

Three things I wish GitHub did: shared codespaces, a collapsible file tree, and branch-scoped tokens for AI agents

GitHub ties a codespace to one person, opens a pull request's file tree at full depth, and lets a token that opens pull requests also rewrite main. The three fixes I want, the workaround I use today, and why AI agents raise the stakes.

Product Building

Forward-deployed engineering: shipping inside a customer's environment

In forward-deployed work the product runs on the customer's data, in their workspace, and done stops meaning what your test suite says. Four rules for the part your tests never reach.

Design × Code

Tolerance vs precision in design: the exception is the foundation

A tile choice turned a contractor's three-centimeter error into a bill. The rule underneath: tolerance is a budget somebody pays, and precision only earns its cost where errors compound.

General

Try the cheap fix first, unless failing it costs you the expensive one

Reversible-first thinking says reach for the cheap option before the heavy one. It is incomplete: a cheap attempt is only cheap if failing it leaves the expensive attempt exactly as available, and exactly as cheap.

Design × Code

3D design vs DevOps: iteration cost decides how many attempts you get

Both disciplines describe a system precisely enough for something else to execute. What separates them is iteration cost: the price of being wrong, which sets how many attempts a career contains.

General

I tore down a dream I'd spent years building

The 3D work was good and the ceiling was real. Tearing it down was not giving up on the dream. It was keeping the dream and demolishing the vehicle that could not carry it.

Product Building

Definition-of-done driven development for teams using AI agents

When agents write the code, your leverage moves to defining what done means, and to building the verification system that checks it. A product builder's guide to the shift.

Design × Code

Designing a resilient house the way I run production systems

No fireplace, one energy source, isolated utility rooms, redundant solar, protected structure. What software and DevOps work did to a house design.

AI Leverage

AI usability testing: probe-ux, an agent that plays your users

Usability testing with a small, un-opinionated agent skill that impersonates any persona and walks any user story against a live product. What it catches, what it can't, and why it's read-only.

Design × Code

Social housing and generative design: AI changed one variable

Social housing failed partly because uniformity was the price of scale. Generative design makes personalization nearly free on paper, which reopens an old question.

AI Leverage

An AI workflow that compounds: run the task through an agent skill

Doing a task by hand improves that one run. Doing it through an agent skill improves every future run: self-documenting, delegable, verifiable, and harder to skip steps in.

Design × Code

3D printing in construction: print the parts, not the house

Mass production serves the median and abandons the rest. The use of 3D printing in building is the low-volume, high-variety part nobody will tool up for, not the whole structure.

AI Leverage

AI for runbooks and docs: the work you never get around to

Most advice treats AI as an accelerator for what you do well. The bigger win is documentation, runbooks and process discipline: the work you always skip.

AI Leverage

Why AI coding agents aren't making your team faster

Teams were promised a 3-10x multiplier and report closer to 10%. The gap isn't the model; it's that most codebases give an agent nothing to check against.

Product Building

When to speak up at work, and when to let it go

Agreeableness at work is a dial, not a personality. Set it too high and you watch avoidable damage happen in silence. Too low and people stop bringing you things.

Product Building

How to prioritize work that can't be undone: the Domino Effect

Impact-effort prioritization assumes your tasks are independent. They aren't. Order work by what it makes cheaper next, and by what still matters after a pivot.

AI Leverage

How to use AI at work without letting it erode your judgment

Handing AI the output is fine. Handing it the decision about what's worth making is how people become interchangeable. A practice for keeping your judgment while the tools get faster.

Design × Code

Affordable housing needs generative design, not just policy

The cost of a house is a stack of numbers, each improvable by a few percent, and they compound once design iteration gets cheap.

Design × Code

Generative design in every design office: the barrier just moved

Generative design stalled in architecture because it needed a programmer on staff. That just changed, and form optimization is still barely explored.

AI Leverage

Plan or prototype first? Just Start beats Just Do It

Some problems can be planned. Others only show their shape once you start building. How to tell which one you face before you plan or prototype, and what AI changed about finding out.

General

Three generations, three economies: land, access, judgment

My grandfather sold what he grew. My father sold what he built, locally. I build software used by people I'll never meet. The bottleneck moved from land to judgment.

Product Building

How to build trust in remote teams: the problem isn't slacking

Remote work removes every ambient signal people used to reassure themselves with. The fix isn't monitoring. It's written process and unambiguous ownership.

General

Why I stopped hunting birds: my sister did the division

I had a defensible position and I was pleased with it. Then someone ran the numbers. On the gap between having a considered view and feeling like you have one.