Would ChatGPT recommend us? Not at first.
We built a door for other companies' AI assistants before a chat window for people. Then we asked ChatGPT whether it would recommend us. It took several goes.
Read the post →
Written for the people who buy software
Short, plain-English notes on delivery, ownership and the evidence behind business software — filed as they happen, kept on the record. No jargon, no war stories for other developers.
We built a door for other companies' AI assistants before a chat window for people. Then we asked ChatGPT whether it would recommend us. It took several goes.
Read the post →
Nine AI models from three makers backed the same wrong score. The practical case for diversity now applies to the AI in your software, and to the AI using it.
Read the post →
A new client asked us to review their code, then to run it. The code was the attack. Why the process held, only just, and the checks any company should make first.
Read the post →
Three 'new best model' launches in three weeks. Why we stay with one maker, switch only within its family, and why how you use AI beats which one you have.
Read the post →
Ben Rodman's work-experience week: a messy brief taken to a working membership system, a client prototype in two days, and a job offer fifteen minutes after his demo.
Read the post →
Two public experiments, and two rules we now build in everywhere, show why a prompt needs tests, a log and a loop, like any other feature.
Read the post →
The rules are real and the fines are large, but the duties follow role and market. A plain-English map of who owes what — and whether it reaches a UK firm.
Read the post →
The law asks less for paperwork than for proof: six things your business must be able to do, and one question that tests them all.
Read the post →
A personal note from our founder: software is data, shaped and governed differently — and most of software development should be configuration, not coding.
Read the post →
Every AI demo works; your Friday-afternoon paperwork is the real exam. How we test AI on your material, your questions and rival models, and share the scores.
Read the post →
Asimov's robot laws were a plot device. Ours run in production: four rules that keep AI useful and honest in business software, with costs and limits stated plainly.
Read the post →
After 25 years building business software, we have launched SDUK Studio: working systems in weeks, at a fixed price, with proof generated by every build.
Read the post →
The headlines say AI killed SaaS. The truth is more useful: what's dying is per-seat rent — and what replaces it matters more than the obituary.
Read the post →
The price of writing software has collapsed. The cost of knowing it's right hasn't moved. What that gap means for anyone buying business software in 2026.
Read the post →