One AI assistant across several functions: under what conditions?
No-code tools make it possible to give several functions access to the same assistant without resolving their differences in vocabulary, constraints or success criteria. The management question is specific: under what conditions does a shared assistant facilitate a handover between functions, rather than conceal their disagreements?
What the literature establishes
Paul R. Carlile (2002, Organization Science) shows, in a study of new product development, that occupational knowledge is embedded in practices and interests that complicate its circulation across functions. Shared objects, such as product representations, can support coordination when they allow people to represent knowledge, examine differences and transform that knowledge. For assistant design, this suggests focusing on handovers between functions, not merely on fluent responses.
Understanding requires connecting contexts
Beth A. Bechky (2003, Organization Science) describes how occupational communities develop shared understanding on a production floor by connecting their knowledge to concrete problems and objects. Sharing is not simply a matter of transmitting identical words: it involves relating what one group says to another group's work. By analogy, a handover note produced by an assistant should explain what a piece of information changes for the receiving function, rather than merely rephrase the original message.
The premature conclusion
One might conclude that an assistant equipped with a common vocabulary would eliminate occupational boundaries. These studies do not justify that conclusion: understanding another department's constraints removes neither divergent interests nor the need for trade-offs. An assistant can help make a disagreement visible; a well-written summary should not be mistaken for operational agreement. (our executive and employee training programmes)
What the evidence does not allow us to promise
Both qualitative studies examine collaborative work, not generative assistants or no-code platforms. They illuminate a coordination mechanism but demonstrate no benefit attributable to an AI assistant. A trial focused solely on response readability would therefore miss the essential question: can the recipient actually carry the work forward?
A practical check in Bâle
Within SHR's programme « Créer ses assistants et agents IA sans coder », a relevant exercise in Basel would be to build a handover assistant linking international R&D, pharmaceutical quality and Rhine logistics, using fictional or authorised life sciences case files. Compare standard handovers with assistant-supported handovers on randomly assigned cases of comparable difficulty, keeping the same human validation requirements. Measure the proportion of files that the receiving function can take on without requesting clarification, using a predefined checklist: understandable status, explicit constraints, and an identified next action and owner. Record critical errors separately: fewer clarification requests would not constitute progress if they concealed a transport constraint or quality status. To go further: explore the Building AI Assistants and Agents Without Coding training in Basel, or browse our executive and employee training programmes in Switzerland.
In pictures: Building AI Assistants and Agents Without Coding in Basel
- building ai assistants and agents without coding
- Basel
- research
- training basel