A different way of building: human direction and GPT
What building this website revealed about human judgement, GPT-assisted implementation, joined-up disciplines and the need to verify what actually works.
Read the articleResearch, case studies and reflections connected by this topic.
What building this website revealed about human judgement, GPT-assisted implementation, joined-up disciplines and the need to verify what actually works.
Read the articleAn investigation into how natural-language collaboration can accelerate practical engineering while retaining architectural authority, verification and human judgement.
Read the studyA practical case study of the WordPress research website: its editorial architecture, accessible presentation, automated draft publishing and observed local test results.
Read the studyHow a graphical diagram workflow became simpler, more accessible and easier to maintain when its output was replaced with semantic HTML and CSS.
Read the articleHow natural-language intentions become bounded engineering changes, with architectural constraints and observable acceptance criteria.
Read the studyWhy rapid AI-assisted implementation still needs explicit responsibility boundaries, stable ownership and human acceptance.
Read the studyAn examination of evidence-led engineering iterations, why working code still needs inspection, and how removing complexity can improve a product.
Read the studyAI-assisted release documentation at Zegeba, using structured product knowledge, screenshots, change review and human validation.
Read the studyHow I conceived, designed and developed determinIQ: a working governed data-intelligence product combining AI-assisted interaction with deterministic analysis, evidence, reporting and privacy controls.
Working implementation Read the studyReflections on using AI to accelerate product engineering while keeping architecture, testing, product acceptance and accountability human-led.
Read the article