How far can AI-driven agentic workflows take software delivery before human intervention is required?

Building autonomous coding agents, evaluating model capabilities and designing human-in-the-loop systems that scale delivery without sacrificing quality.

Agentic coding is reshaping how software is built. Understanding where autonomy works and where it breaks is essential to leading engineering teams through this transition.

Can a deliberately small, focused product earn attention in a market dominated by feature-rich platforms?

Designing and building WordFloor — a constrained writing tool — as an exercise in product discipline, focused scope and sustainable independent software.

WordFloor tests whether a single-person product can compete through craft, focus and a clear point of view rather than feature volume.

How can AI improve product discovery in ecommerce beyond traditional search and faceted navigation?

Exploring semantic search, personalisation and LLM-driven product discovery within the constraints of real ecommerce catalogues and retail operations.

Ecommerce search is a high-impact applied-AI domain where small improvements in discovery quality directly affect revenue, conversion and customer satisfaction.