AI: speeding up one stage without blocking the next
In SHR — Swiss Human Resources’ programme “Managing in the Age of Artificial Intelligence”, coordination deserves a question distinct from individual productivity. How can a manager check that a stage accelerated by AI does not shift the waiting time to the next team?
What the literature establishes
Thomas W. Malone and Kevin Crowston (1994, ACM Computing Surveys) propose understanding coordination as the management of dependencies between activities. Their synthesis distinguishes, among other things, shared-resource problems and relationships between activities that produce an output and those that use it. Applied to AI, this framework invites managers to examine not only how quickly something is produced, but also the conditions under which another department can use it.
The mechanism: making handovers work
Gerardo A. Okhuysen and Beth A. Bechky (2009, Academy of Management Annals) identify three integrating conditions for coordination: accountability, predictability and common understanding. Their review explains how mechanisms such as routines, meetings and plans can contribute to these conditions. An AI-generated summary therefore does not, by itself, establish who should take it forward, when or according to which criteria.
The premature conclusion
One might conclude that every handover simply needs standardising or an additional approval step. These studies do not support prescribing that solution in every case: an extra procedure can also become another source of waiting. The aim is to make the dependencies that matter explicit, not to multiply checks. (our executive and employee training programmes)
What these studies do not demonstrate
Both articles are conceptual and review contributions that predate the current spread of generative AI. They establish neither that AI causes bottlenecks nor that a particular coordination mechanism eliminates them. Measuring only document preparation time would, however, leave out the time it spends waiting, any return for revision and its actual use by the receiving department.
A practical check in Sion
In Sion, a manager could select a single non-critical workflow: preparing a maintenance file in hydropower, an administrative handover in healthcare, coordination between cellar operations and sales in wine production, or a handover between reservations and reception in Alpine tourism. Over two periods of equal length, before and during a supervised AI trial, the team would record each comparable file’s initial receipt, handover and acceptance as usable by the next department, without exporting sensitive data. The main measure would be the median time from initial receipt to acceptance, alongside the proportion of files returned for additional information and the number still waiting. This comparison alone would not establish a causal effect of AI, but it would allow the team to check whether the claimed gain also appears beyond the role producing the output. To go further: explore the Managing in the Age of Artificial Intelligence training in Sion in the canton of Valais, or browse our executive and employee training programmes in Switzerland.
In pictures: Managing in the Age of Artificial Intelligence in Sion
- managing in the age of artificial intelligence
- Sion
- research
- training sion