Consulting

Who I work with

Two kinds of organisation.

• Companies of one hundred to a thousand people, where the owner or the board shapes processes directly and can decide on a direction within a short time. Here the path from decision to first results is the shortest.
• Enterprise organisations, often regulated, for which certainty is a precondition: knowing where the data is, who is accountable for a solution, and what happens when the model vendor changes the rules.

My counterpart in the conversation is the board. The IT department is an ally without which no deployment survives in operation. I involve security and compliance at the start of the work, not at its end.

Where I start

With strategic intent. AI adoption that is meant to reach the whole organisation requires decisions at board level: on objectives, on the boundaries of acceptable risk, on accountability, and on investment in the teams’ competence. Without that decision, deployments remain a series of independent initiatives that can be neither scaled nor accounted for.

The strategic board session. Half a day of workshop in which board members build a working solution themselves and, on that basis, assess what the technology genuinely changes in their organisation. The outcome: a shared understanding of possibilities and limits, a preliminary list of processes with the greatest potential, and a decision on the direction and scope of further work.

What I do next

• Diagnosis. A map of candidate processes, a baseline measurement, an assessment of technical, regulatory and data readiness. An inventory of solutions already in use in the organisation, including those introduced without the board’s knowledge.
• The governance foundation. Usage policies, a register of AI systems, risk classification, ownership, rules for human oversight, data architecture and location. The stage that decides whether the organisation will be able to scale deployments beyond the pilot.
• A pilot with a measurable result. Six to eight weeks on a selected process, a defined metric, an economic case. Scaling follows only once the result is confirmed.
• Team competence. A programme tailored to roles, covering work with the tools, safe handling of data, and the design of human–machine collaboration. The competence stays in the organisation, not with the vendor.
• Direction review. A regular assessment: what produced results, what has changed in the technology since the last review, and where the next step lies.

How I work: seven commitments

They are the basis of every engagement, and you are entitled to hold me to them.

1. I build managed AI. Every solution has an owner, documentation, a described data flow and planned maintenance. Solutions without these elements are not built, even on request.
2. Data stays in the organisation. I work on models running in the client’s infrastructure, in an agreed processing region. Any architecture that moves data outside the controlled area requires informed, written consent.
3. I prove it with numbers. A baseline measurement at the start, a measurement after the change at the end. Solutions that do not pay back within a reasonable horizon are not built.
4. I advise on technology choices and do not earn from them. I take no commission from vendors, because I would then stop being an adviser and become a sales channel.
5. I recommend AI only where it is the right answer. Some problems are solved by classic automation, some need a process correction, and some need nothing but a board decision. I say so plainly, even when it means a smaller engagement.
6. I do not leave the organisation alone. The technology changes faster than the planning cycle. I return at an agreed rhythm so that the organisation keeps pace with the change and the solutions stay current.
7. I leave capability, not dependency. I document solutions so that the client’s team can take them over completely. I do not build an engagement model based on dependence on the adviser.