Your AI vendor can quickly understand what you can no longer do without them. That is the moment you realise you did not choose a tool — you accepted a dependency. This moment usually comes at renewal. We intervene before.
Most AI vendor selections are conducted by technical teams on performance, cost and integration criteria. These criteria are necessary. They are insufficient.
The missing criteria are strategic: what is the exit cost in 3 years? Which data will the vendor use for its models? Which contractual clauses can be modified unilaterally? These questions determine your situation in 3 to 5 years, not 3 to 5 months.
An AI vendor is not a third party. It is an extension of your organisation: your data, your decisions, your critical processes flow through its systems. What you entrust to it does not always come back intact if you leave.
The more you outsource, the more you gain in execution speed. The more you gain in speed, the more you lose in control. This trade-off is unavoidable. It must be arbitrated consciously, not endured by default.
What is the contract exit procedure? Within what timeframe? At what cost? What happens to your data and fine-tunings if you leave? A vendor that cannot answer clearly does not want you to leave.
Does your data feed the vendor's model training? Under what conditions? Is there an explicit non-training clause? If this clause is not in the contract, it does not exist.
AI vendors can modify their models, pricing, usage limits or contractual conditions — sometimes without notice compatible with your strategy. Which contract elements can be modified unilaterally?
Your prompts, fine-tunings, evaluations, workflows — are they portable or captive to the vendor's ecosystem?
We assess what each option truly commits you to in terms of control, exit cost and competitive position.
let's talk