Bridging the Data Gap: A Semantic Translation Layer for UK Poverty Data Ponderings on how to bridge the data gap. Exploring questions, semantic layers, TRE's and feedback loops.
Building Open Recommendations A full breakdown of how I built Open Recommendations, the inner workings, the choices I made, including provider agnostic AI, strict data structures, deciding whats a recommendation and whats just an observation, RAG, and weighting community feedback.
My Memory - A startup idea for 2026 One of a number of ideas I'm thinking about ahead of 2026. The personal context passport lets people carry a portable, user-owned profile they can share on their terms - from low-stakes convenience (travel, fitness) and evolving into granular, time-limited permissions for high-stakes public services
Predictions for 2026 (or: What I'm Actually Thinking About) The Predictions for 2026 that no-one asked for. Some of them are real, some are more just hope, most of them are firmly tongue in cheek. Also I set out 8 ideas I'm thinking about when it comes to tech and AI infrastructure.
People as Code 4 - Entropic Systems Embracing Entropy as a concept for design. Turning from the theoretical to the practical. Observability, not measurement. Questions as interfaces, data as plural
Lessons learned from working together Data for Action worked with ClientEarth for around 18 months. We recently held a retrospective to explore what we all learned about the work. This document looks to summarise the lessons we can learn from.
People as Code People as Code imagines a world built like open source, modular, participatory, and transparent, where people help write policy, foundations build platforms, and data becomes a shared resource.
Grant funding for digital and data in the UK charity sector - a year on have things changed and perhaps how they could change for the better