AI assistants: productivity or a jagged frontier?
Building an assistant without code makes experimentation accessible. It does not remove the need to assess the task and verify the output.
An experimental result
Shakked Noy and Whitney Zhang (2023, Science) randomly assign professionals to writing tasks with or without generative AI. Assisted participants finish faster and receive higher average ratings. The tasks are short and standardised, not complete organisational processes.
The jagged frontier
Fabrizio Dell’Acqua and colleagues (2023, Harvard Business School Working Paper) study consultants: AI helps substantially within its capability but may worsen answers beyond it. A working paper is not definitive evidence.
Why no-code is not enough
Easy configuration shifts risk towards instruction design, source quality and output control. A fluent assistant is not necessarily a reliable one. (our executive and employee training programmes)
A transferable capability
Managerial value lies less in a prompting formula than in decomposing the task, defining quality criteria and matching verification effort to risk.
A practical check in Zurich
Test the assistant on ten resolved cases, including three edge cases. Compare outputs with the reference and record plausible but false answers before live use. To go further: explore the Building AI Assistants and Agents Without Coding training in 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