IA on the job: does a proven solution still fit?
An assistant drawing on internal files can quickly bring a previously successful solution back into use. The management question is precise: how can we check that the conditions behind its success still hold before reusing it?
What the literature establishes
James G. March (1991, Organization Science) distinguishes the exploitation of existing knowledge from the exploration of new possibilities. His theoretical analysis shows how the more immediate and predictable returns from exploitation can lead an organisation to give it excessive weight. Applied to AI, this framework invites us to examine not only how easily an old solution can be retrieved, but also how much room remains for seeking another response.
When success narrows the search
Daniel A. Levinthal and James G. March (1993, Strategic Management Journal) analyse the myopias of learning, including the tendency to favour nearby effects and overlook failures. An organisation can therefore become more proficient in its established practices without becoming better at recognising when they no longer fit. An assistant that mainly resurfaces files regarded as exemplary could reinforce this mechanism; this is a proposed application, not a finding from that study about AI.
The premature conclusion
One might conclude that AI should always be asked for an unprecedented solution. That would confuse exploration with novelty of wording: a different answer is not necessarily a better fit. The task is instead to make explicit the conditions under which a precedent can be transferred, then establish whether those conditions hold in the current case. (our executive and employee training programmes)
What these studies do not establish
These articles offer theoretical analyses of organisational learning, not evaluations of generative assistants in companies. They therefore establish neither the frequency of inappropriate reuse nor the effectiveness of an instruction asking a model to challenge precedents. In practice, a file archived as a success may also leave undocumented the commercial, regulatory or relationship conditions that made it possible.
A practical check in Vevey
Within SHR’s « IA on the job » programme, an exercise in Vevey could examine the reuse of responses to supplier requests, against the local backdrop of global food businesses, group headquarters and Riviera SMEs. For two weeks, every proposal based on a precedent would include a verifiable condition of validity — the relevant market, a quality requirement or a contractual provision — and a document from the current case against which to check it. The responsible business manager would measure the proportion of reused proposals whose condition was actually checked before sending, then record those amended because the context had changed. This record would verify a disciplined approach to transferring solutions, without claiming on its own to demonstrate a performance gain. To go further: explore the AI on the Job training in Vevey in the canton of Vaud, or browse our executive and employee training programmes in Switzerland.
In pictures: AI on the Job in Vevey
- ai on the job
- Vevey
- research
- training vevey