AI strategy, decided with you
Senior counsel on where artificial intelligence belongs in your organisation — and where it does not. Independent of what we build, so the advice is a recommendation rather than a sales pitch.
Book a free callThe three decisions that are expensive to reverse.
Advisory here is not a strategy document. It is a short list of decisions, each with its trade-off, its cost and what it would take to undo.
Where AI belongs — and where it does not
An honest read on which parts of your operation AI genuinely improves, which it would make slower or riskier, and what is better left alone. Saying no is part of the service.
Build, buy or wait
Which capabilities are worth owning, which to licence, and which to postpone until the technology settles — with the lock-in risk of each named before you commit.
Governance — who decides, and who answers
Who signs off on an AI-assisted decision, what gets logged, and who is accountable when the system is wrong. Most organisations discover they have no answer at the worst possible moment.
Where does our data actually go?
This is the question leadership asks before any other, and the one that quietly stalls AI projects. It has concrete engineering answers rather than reassurances — and which answer is right for you is a decision, not a default.
What never leaves the building
Classify before you connect. Some data can safely reach a third-party model, some may only reach a model you control, and some should never appear in a prompt at all. Most organisations hold all three and have never separated them.
Anonymise before you send
Names, client identifiers and personal data stripped or replaced at the boundary, before a request ever reaches an external provider — so the useful signal goes out and the identifying detail stays with you. Done properly it costs nothing in quality.
Open-source AI in the cloud, without surrendering your data
Most builds use the strongest model available for the job. When data protection changes that calculus, this is the answer — and it needs no GPU of your own: open-weight models served from Swiss and European clouds by providers that neither log your requests nor train on them. Current, capable, and your data left alone. For the rare workload that cannot leave a perimeter you control at all, we run the model on our infrastructure or on yours.
A standing counsel, not a deliverable.
AI moves faster than an annual strategy cycle: models are deprecated inside a year and what was impossible last quarter ships this one. Advisory is therefore a relationship, reviewed as the ground shifts.
Orientation
We learn the organisation — its constraints, its politics, its ambitions and what has already been tried. No recommendation is worth anything without this.
Decisions, documented
Each recommendation written down with its reasoning, so it can be challenged, and so it still makes sense to whoever reads it a year from now.
Standing counsel
A retained sounding board as the ground moves — a vendor changes terms, a model is retired, a new opportunity appears. You get a call, not a proposal.
Understanding digital transformation services.
How is this different from the process audit?
The audit examines your PROCESSES — fixed scope, ten days, mapped and quantified, ending in a prioritised roadmap. Advisory addresses the DECISIONS above that: whether to pursue AI at all in a given area, what to own versus licence, and what your data and governance must look like first. Many clients take the audit alone. Advisory suits those making commitments large enough that the wrong one is costly.
Do we have to buy anything from you afterwards?
No, and the advice is written on that basis. We will tell you when an off-the-shelf tool beats anything custom, when the honest answer is to wait, and when the work should go to someone else. Advisory that only ever concludes "hire us to build it" is a sales call with an invoice attached.
What if the answer is that we should not use AI?
Then that is what you will hear, with the reasoning. A great deal of what is sold as AI today is a worse version of a spreadsheet, a clearer process or a decision somebody needs to make. Finding that out in an advisory conversation is considerably cheaper than finding it out in production.
Do we need technical expertise internally?
No. The conversation is held in business terms, and the technical detail sits underneath for whoever wants it. Where your team does need to grow into something, we say so plainly rather than leaving a dependency behind.
Do you only work with Swiss organisations?
No. We are based in Lausanne and Geneva and much of the work is with organisations across Suisse romande, but GMG grew out of work for international federations and event organisers — so multi-country, multilingual operations are familiar ground. Anything we host stays in Switzerland.
Not sure whether AI is the answer?
That is exactly the conversation to have first. The opening 30-minute call is free and carries no commitment.