AI transformation · Processes · Agents · Governance
I help boards move from ad hoc AI use and isolated experiments to a coherent transformation model: the right use cases, redesigned processes, secure data, pilots, adoption and measurable outcomes.
I don't sell licences or technology for its own sake. I start with the business problem and lead the change from diagnosis to implementation.
Measure of value
If we cannot point to a measurable impact, I don't start the project.
01 / Reality
Employees use public tools. Departments buy more licences. Assistants, automations and pilots appear. Company data ends up in systems that nobody has centrally approved.
That is AI transformation too — just an uncontrolled one.
Another licence will not solve this.
02 / Diagnosis
AI transformation starts before the choice of technology.
The most expensive mistake is automating a process the company does not need at all.
03 / How I help
I establish where AI can build an advantage, increase revenue, lower cost or improve the quality of decisions.
The strategy also defines what the company will not do, what risk it accepts and where people must remain in control.
I analyse processes, decision points, repetitive tasks and the places where knowledge gets lost.
The result is a portfolio of use cases assessed by:
I don't bolt AI onto an existing procedure without checking whether the procedure itself makes sense.
I design the target way of working: what the person does, what the system does, when approval is needed and what happens in an exception.
I help define the solution's role, its scope of action, the data it needs, its permissions and the security conditions.
This can include:
I look for applications directly linked to commercial results:
I define the rules for using AI:
Technology only works when people are genuinely able and willing to use it. That is why capabilities, communication and adoption are part of the implementation — not an add-on at the end of the project.
04 / The right tool
01
For a stable, repetitive process with clear rules.
Example: moving data, generating a report, sending notifications.
02
When a person needs to analyse, create or find information faster.
Example: preparing a recommendation, summarising documents, working with a knowledge base.
03
When a system needs to carry out a multi-step task on its own, use tools and respond to a changing context.
Example: gathering data, analysing it, preparing an action and executing it subject to human approval.
04
When a decision has a high financial, legal or reputational impact, or concerns another person.
AI can prepare the information and a recommendation. Responsibility stays with a person.
05 / Security
An agent with access to CRM, email, documents or company systems is no longer just a chatbot. It becomes a new participant in the process.
That is why, from the design stage, I take into account:
Security is not meant to block innovation. It is meant to let it scale safely.
06 / How I work
I check where AI is already used, what tools the company has, what data is available and where the greatest business potential lies.
I assess use cases by value, feasibility, risk and time to impact.
I define the target way of working, the roles of people and AI, data, permissions, measures and exceptions.
We launch a limited solution with a clearly defined user group, a baseline and success criteria.
After the pilot we make a deliberate decision: scale, improve, stop or choose a different use case.
We build the rules, capabilities, accountability and operating model that allow further solutions to be rolled out faster and more safely.
07 / What you get
Depending on the scope, you receive:
I don't leave the company with a catalogue of tools. I leave it with a decision on what to implement, why, how, and how to measure the result.
08 / Ways of working
Recommended starting point
2–3 weeks
For a board that wants to understand its starting point, bring order to experiments and choose a few use cases with the greatest value.
Outcome:
02
For a company that wants to redesign a specific process and move from concept to a measurable pilot.
Outcome:
03
For an organisation already running several initiatives that needs a shared strategy, governance, operating model and management of its implementation portfolio.
I can work as an adviser to the board, as transformation programme lead, or temporarily take responsibility for connecting the business, data, IT and implementation partners.
Case study · International services platform
Programme outcome
Results of the entire loyalty programme using data, automation and AI mechanisms.
09 / Who it is for
I am not a software house or a technology vendor. Nor will I be a good partner if all you need is prompt training or a chatbot that looks good in a presentation.
I can, however, define the problem, design the model, prepare requirements, help choose technology and partners, and lead the implementation on the business side.
No. Technology is an important part of the solution, but the business should own the outcome. Transformation concerns processes, data, accountability, capabilities and the way decisions are made.
I don't operate as a software house. I define the use case, business model, process, requirements, security rules and measures. I can work with internal IT or a chosen technology partner and lead the project on the business side.
Yes. Assessing the data is part of the diagnosis. Sometimes the first project should not be an agent, but getting knowledge, access and data ownership in order.
I look for the intersection of four things: a significant business problem, available data, real feasibility and a result that can be measured in a reasonable time.
Before the pilot I establish a baseline: cost, time, number of errors, conversion, revenue or another appropriate measure. Only then do I compare the solution's result with the starting point.
No. Some processes should be simplified or removed. In others the best solution will be plain automation rather than AI.
They can be, if they have limited permissions, work on the right data, their actions are logged and high-risk decisions require human approval.
Yes, but a workshop should lead to a decision. Its outcome can be a shared AI ambition, a process map, a choice of use cases or a pilot preparation — not just inspiration.
You need one well-chosen change whose result can be measured — and then a model that lets you implement the next ones.
