Most writing about AI development is either vendor marketing or academic papers. Neither is much help when you're deciding whether to build something, working out what it will cost, or trying to figure out why your current system isn't working. We write about projects we shipped, and we use the numbers we measured on them.
We publish on three themes. Cost and scoping: what AI projects really cost in 2026, how to work out the return on an automation project, and when buying makes more sense than building. Architecture: retrieval design, evaluation harnesses, and the choices that decide whether a demo can become a system paying customers rely on. Operations: life after launch, where token costs creep in, and how output quality drifts over time.
Everything here comes from work for real clients. When we quote a figure, such as a 64% cost reduction or a 95% accuracy rate, it was measured on a system running in production.
Want a number for your own project? We build AI agents, AI automation, RAG systems, mobile apps and web apps for clients across the United States and Australia. Book a free 30-minute discovery call and we will send a written fixed-price proposal within 3 business days.