AI on the Business — Organisational Inquiry Prompt v1.0

This prompt is designed to help you use generative AI to examine a piece of organisational work before deciding how it should be improved or whether any part of it should be automated.

Choose a recurring piece of work that you know reasonably well. Describe it to your AI system and, where appropriate and permitted, provide examples of the documents, information or other evidence associated with it.

Then use the following prompt.

I want to understand how this work actually operates before considering how it might be improved.

Your role at this stage is to help me investigate and understand the work, not to redesign it or recommend technology.

Start by questioning me about the work. Develop your understanding progressively rather than asking me for everything at once.

Explore:

• what triggers the work and what outcome it is intended to produce;
• who is involved and what each person contributes;
• what information is required, where it originates and how it moves;
• which systems, documents, spreadsheets and other tools are used;
• where responsibility passes between people, teams or systems;
• where work waits or is delayed;
• where information is entered, checked, copied or transformed;
• where rework, duplication or repeated checking occurs;
• what happens when information is missing or incorrect;
• what exceptions occur and how people deal with them;
• what workarounds people have developed;
• what controls and approvals exist and why;
• where professional or managerial judgement is required;
• where the documented or intended process differs from what happens in practice.

Do not assume that my description is complete or correct. Question ambiguities and apparent contradictions.

As we proceed, distinguish between:

Evidence — something we can establish from records, observation, examples or other sources.

Interpretation — an explanation we are currently giving to that evidence.

Hypothesis — a possible explanation that needs further investigation.

Do not convert an interpretation or hypothesis into a fact simply because it appears plausible.

Identify things that I appear to be assuming but have not demonstrated. Where appropriate, tell me what additional evidence would help us establish what actually happens.

If something cannot be established from the information available, say so.

Do not recommend improvements, automation or AI applications during this stage. If you notice an apparent improvement opportunity, record it as something to investigate rather than proposing a solution.

Continue questioning me until we have a sufficiently useful account of how the work currently operates.

Then produce a structured summary containing:

1. the purpose and intended outcome of the work;
2. how the work appears to operate in practice;
3. the people, systems and information involved;
4. significant handoffs and dependencies;
5. waiting, duplication, rework and workarounds identified;
6. important controls and judgement points;
7. differences between the intended process and actual practice;
8. unresolved questions;
9. assumptions that still need checking;
10. additional evidence that would improve our understanding.

End there.

Do not propose solutions until I explicitly ask you to move from understanding the work to considering possible interventions.

When you are ready to examine possible interventions

Only after you have checked the description against appropriate evidence and, where necessary, with the people involved in the work, continue with:

Now help me examine this work critically. For each apparent problem or improvement opportunity, separate the evidence from our interpretation of it and identify any assumptions that still need checking. Consider the underlying cause rather than immediately treating the visible symptom as the problem.

For each opportunity, consider whether an appropriate intervention might involve removing work, simplifying it, standardising it, changing responsibilities, improving information, training, system configuration, conventional automation, AI assistance or AI agency. Do not assume that technology, automation or AI is the preferred intervention.

Where there is insufficient evidence to recommend an intervention, identify what we need to establish before deciding what to do.

A note on using the prompt

The quality of the analysis depends on the quality of the evidence available to it. An AI system does not observe your organisation simply because you describe a process to it.

Use its questions to investigate the work. Talk to the people who perform it, examine examples and records, and check assumptions against what actually happens.

AI can accelerate organisational inquiry; it cannot substitute for it.

Related reading

This prompt is the practical companion to Why AI should work on the business, not just in it.