AI assistants: should Zurich teams keep expanding their toolkits?
In Zurich’s financial services, insurance and technology businesses, as well as the European headquarters of international groups, every function may want its own AI assistant. The management question is specific: does expanding the catalogue help employees work, or does it impose an additional decision before each task?
What the literature establishes
Sheena S. Iyengar and Mark R. Lepper (2000, Journal of Personality and Social Psychology) showed, in experimental settings involving consumption and task performance, that a broader selection could attract more interest without encouraging action. Some of their experiments associate more limited choice with greater engagement or higher satisfaction. These findings establish an important possibility: making more options available does not necessarily make decisions easier.
A second study that cautions against generalisation
Benjamin Scheibehenne, Rainer Greifeneder and Peter M. Todd (2010, Journal of Consumer Research) revisited this question in a meta-analysis. They found an average choice-overload effect close to zero, alongside substantial variation across studies. Difficulty therefore cannot be inferred from the number of options alone: the conditions under which people choose matter.
The premature conclusion
One might conclude that an assistant catalogue should always be kept to a minimum; the evidence does not justify that rule. A specialist may need distinct tools for distinct tasks, while an occasional user may struggle to distinguish assistants with similar descriptions. The hypothesis to test therefore concerns the clarity of intended uses as much as the size of the catalogue. (our executive and employee training programmes)
The limits of the evidence and of current practice
These studies do not examine AI agents embedded in business processes and establish no optimal number of assistants. Counting the tools created or how often they are opened is likewise insufficient to determine whether employees find the right one for their task. In SHR — Swiss Human Resources’ programme « Créer ses assistants et agents IA sans coder », this distinction provides a design question: how can each assistant’s scope be made identifiable before use?
A practical check in Zurich
Within a Zurich team, randomly assign users to two presentations of the same assistants: a flat catalogue and a task-based entry point, leaving the tools and access rights unchanged. Using non-sensitive work scenarios, record the time taken to choose and the proportion of initial choices that match the scope defined beforehand by the relevant business leads. A presentation should be considered preferable only if it makes selection easier without reducing the quality of the resulting work, assessed separately using a common rubric. To go further: explore the Building AI Assistants and Agents Without Coding training in Zurich in the canton of Zurich, or browse our executive and employee training programmes in Switzerland.
In pictures: Building AI Assistants and Agents Without Coding in Zurich
- building ai assistants and agents without coding
- Zurich
- research
- training zurich