Systems · Point of view
The second time
should be cheaper.
Component libraries, clear build rules and automation reduce repeated work. Here is how I connect them while keeping people responsible for the important decisions.
What is wrong with project-by-project web delivery?
When familiar components and workflows are rebuilt for each project, clients pay for repetition and delivery depends on more manual work.
Reusable foundations reduce that repeated effort and make quality easier to maintain.
What is a delivery system?
A delivery system is a reusable set of decisions: a token layer, component library, content model, build pipeline and clear rules for when a human intervenes.
The output can vary. The expensive decisions do not have to.
Where does AI actually fit?
AI output is easier to review when it follows a defined structure. I build the structure by hand—sections, tokens, motion rules and performance budgets—then use generation for content, configuration and assembly inside those constraints.
The output still needs validation. Clear constraints make that review more consistent.
One foundation. Different outputs.
Build once.
Reuse the decisions.
Keep each project distinct. Reuse the tested foundation to reduce repeat work and make updates more consistent.
Craft, made repeatable.
Where should humans stay?
Approval gates, judgement calls and anything where being wrong is expensive. Automating the typing is different from automating the decision.
The system should remove repetition without removing accountability.
What am I building now?
PROMTURE is developing internal infrastructure for AI-assisted digital delivery: structured intake, creative direction, asset generation, component-led builds, automated quality checks and controlled deployment.
The goal is not to remove craft. It is to make craft repeatable.
A useful system starts with the right constraint.