AI: can you change providers without losing your know-how?
Choosing an AI solution often involves more than buying a service: teams gradually embed their examples, criteria and methods in it. For managers, the precise question is whether they can leave the provider without having to rebuild the way their team works.
What the literature establishes
Oliver E. Williamson (1981, American Journal of Sociology) presents a framework in which asset specificity helps explain how transactions are governed. When an investment loses value outside a particular relationship, switching partners becomes more costly. Applied to AI, this framework invites scrutiny of provider-specific adaptations, rather than the licence price alone.
Keeping the files is not enough
Wesley M. Cohen and Daniel A. Levinthal (1990, Administrative Science Quarterly) describe absorptive capacity: recognising the value of external knowledge, assimilating it and applying it, drawing notably on prior knowledge. Their analysis suggests that accessing a technology and knowing how to benefit from it are different matters. For a team using AI, retaining instructions and examples therefore does not, by itself, guarantee the ability to reuse them elsewhere.
The premature conclusion
One might conclude that all customisation or long-term commitments to a provider should be avoided. That would confuse a useful investment with an inevitably harmful dependency: a specific adaptation can also improve the service delivered. The managerial task is to make that dependency explicit and preserve what needs to remain transferable. (our executive and employee training programmes)
What these studies cannot promise
These articles test neither today's generative AI platforms nor the costs of migrating between them. They offer analytical frameworks, not a risk estimate for a business in Fribourg. An export clause or a sales demonstration is therefore insufficient to establish whether a team can resume its work with another tool.
A practical check in Fribourg
In Fribourg, an exercise in SHR's « Manager à l'ère de l'intelligence artificielle » programme could examine the transfer of a non-sensitive use case: preparing a handover sheet for an agri-food business, a bilingual industrial company or a university service. The team would assemble a provider-independent package containing instructions, a French–German glossary, authorised examples and acceptance criteria, then attempt to reproduce the work in another approved solution. The measure would be the proportion of cases meeting the criteria established before the transfer, alongside reconfiguration time and the number of items that could not be exported. This test would document an actual ability to exit, rather than a merely contractual possibility. To go further: explore the Managing in the Age of Artificial Intelligence training in Fribourg in the canton of Fribourg, or browse our executive and employee training programmes in Switzerland.
In pictures: Managing in the Age of Artificial Intelligence in Fribourg
- managing in the age of artificial intelligence
- Fribourg
- research
- training fribourg