Blog
Beyond keys: The rise of semantic and prefix caching for LLMs
An exploration of how new caching techniques from Redis, OpenAI, and Claude are tackling the expensive problem of repetitive LLM calls.
Onboarding AI with READMEs and quality gates
A practical approach to structuring project documentation using READMEs for context and automated checks as 'Quality Gates' to improve AI-assisted development.
Simon Willison on the lethal trifecta and MCP security
A look at Simon Willison's latest talk on AI security, focusing on his 'lethal trifecta' concept and the risks of the Model Context Protocol (MCP).
A periodic table for system design
An exploration of Joy Arulraj's 'periodic table' of system design principles, with some reflections on its practical use.
A deeper look at AI's impact through the SPACE framework
A reflection on a study that uses the SPACE framework to measure the real-world impact of AI on developer productivity, satisfaction, and collaboration.
A practical framework for trusting your gut
Phil McKinney offers a three-step framework to turn unreliable gut feelings into a systematic and reliable tool for decision-making.
An oral history of how ChatGPT disrupted the field of NLP
A summary of a Quanta Magazine article that documents the shift in Natural Language Processing (NLP) following the rise of LLMs.
A practical guide to setting company policies
A summary of and commentary on Stay SaaSy's excellent article about the lifecycle of creating, managing, and retiring company policies.
Leadership co-processing with LLMs
A look at how Large Language Models can be used as a thinking partner for leadership tasks, inspired by an article from James Stanier.
Experimenting with Kilo Code and the Model Context Protocol
A look at my first experiences with Kilo Code, an open-source alternative to Cursor, and the process of getting the Playwright MCP working.
Putting the developer at the centre of platform engineering
A look at how the practice of continuous discovery helps build platforms that developers actually want to use.
Women as technology
A reflection on the 'Women as Technology' exhibition, exploring the intersection of gender, stereotypes, and technological development.
The bitter lesson for organisations
Ethan Mollick's article on 'The Bitter Lesson' suggests that brute-force AI may outperform human-designed processes, but where do humans still hold the advantage?
The cost of being wrong
A reflection on Jack Vanlightly's article about how the low cost of failure in software should encourage decisive action.
A refreshing, back-to-basics approach to coding with AI
A look at antirez's argument for using LLMs as direct, human-controlled add-ons, avoiding agents and retaining full control over the development process.
People, principles, process, then product
A reflection on Mike Fisher's framework that prioritizes people, principles, and process before focusing on the product itself.