Insights
Insights
Writing on Claude, on what AI changes in a business, and on the work of putting your own data to use.
The data map is the worked example that goes with chapter three, built from an invented business.
21 August 2026
Chapter seven of the AI strategy series. There are three ways a business can work with a machine: assignment, where it does a defined piece of work and gives the result back; augmentation, where it works alongside a person who stays involved throughout; and agency, where it acts, and keeps acting, without anyone checking each action. Risk rises with the freedom, because the freedom decides how far a mistake can travel before someone notices, so the business has to state which mode it is using before the work is handed over.
Read →
20 August 2026
Chapter six of the AI strategy series. Judgement is the decision that commits the business to a course of action, and it disposes of something that could have been spent elsewhere, so it cannot usually be reversed without cost. The fourth map records what was committed to, who was permitted to take it, what was expected to follow from it, where it is recorded, and whether the consequence is ever set against the expectation.
Read →
20 August 2026
Chapter five of the AI strategy series. Information says that a figure moved; knowledge is the explanation of why it moved, and it is sound or unsound relative to the evidence for it rather than accurate or inaccurate relative to the data. The third map records each explanation the unit acts on, the evidence behind it, whether it came from a defined process or from one person, where it is written down, and whether it can be replaced by seniority alone or only by evidence.
Read →
19 August 2026
Chapter four of the AI strategy series. Information is data that has been cleaned, organised and put into context, so the second map records the processing rather than the facts: what is done to the data, whether that is a defined process or a habit in one person’s head, who can change the result, and whether the route back to the source data can be traced at all.
Read →
18 August 2026
Chapter three of the AI strategy series. Data is raw facts with no context attached, and a report, a dashboard or a summary of the month is none of those. The map records where each fact comes from, what format it is in, where it is stored and who can access it, and the discovery is the gap between the facts the unit has recorded and the facts its objectives require.
Read →
17 August 2026
Chapter two of the AI strategy series. Nothing can be improved unless it can be measured, and nothing can be measured unless it can be observed. Almost everything a business does is visible to somebody, and very little of it has been written down, so the audit maps the people, the processes and the handovers between them, and marks what was inferred rather than seen.
Read →
17 August 2026
Chapter one of the AI strategy series. Most AI work in a business starts with the technology, which is the wrong place. Begin instead with the one thing the business is trying to bring about, then the discrete steps that realise it, because a business can meet every objective it set and still miss what it actually wanted.
Read →
4 August 2026
Article 50 became applicable on 2 August 2026. The claim that every piece of AI-generated content must now be labelled, on pain of a €15 million fine, is a significant overstatement. Four duties are in force, two of them are likely to touch you, and one exemption covers most of the work a business does with AI.
Read →
27 July 2026
Usage numbers tell you how much AI is being used but they do not tell you if anyone is using it well and improving. The Compliance API gives you the final outputs but not the working, so a complete fluency assessment takes three lenses: the outcome against the AI’s last draft, the session trace, and the person’s own account of what they did with it.
Read →
26 July 2026
The most impressive AI projects deliver the least, because they start with the clever thing and go looking for somewhere to put it. Map how the team actually works first, build only where an agent genuinely helps, and treat going live as the start of the loop.
Read →
25 July 2026
These things happened to real companies. In nearly every case the AI did what it was asked, and the harm came from an ordinary control nobody had put in place. Seven real cases, and how to stop the same thing happening to your business.
Read →
23 July 2026
IDC describes AI agents arriving in three waves, from a tool one person uses to AI woven through how the business runs. Most customer experience teams are renting the layer they run on. Whether to keep renting a third party’s stack that holds your data, or own the data and run the model directly on it, is the choice this piece is about.
Read →