How to find the work in your business that AI should actually do
A practical way for small business owners to spot which work is worth handing to AI: the three-box test, a one-week audit, and how to pick the first task.
Read →Insights
Notes on building and operating AI
Practical thinking on choosing, delivering, recovering and assuring AI products — written for the people who have to make them work.
How to measure whether AI is actually paying off, without a data team
A plain method to tell whether your AI spend is paying off: baseline the before, measure one task in pounds and hours, count hidden costs, set a review date.
Read →What AI assurance means before production
Before an AI product goes live, assurance is the evidence that it is reliable, controlled, secure and operating within defined boundaries — not a single sign-off.
Read →What is agentic programme recovery?
Agentic programme recovery is the work of diagnosing why an AI agent programme is stuck, resetting it, and taking hands-on accountability for getting it to production.
Read →How to choose your first AI use case
A practical way to pick a first AI use case: score each candidate on business value, user impact, data readiness, feasibility and risk, then start where they align.
Read →