Start with one piece of work
A practical way to investigate one recurring piece of work before deciding whether AI or another intervention is appropriate.
Using AI to understand an organisation can sound like a large undertaking. It does not need to be. Start with one recurring piece of work. Preparing a quotation. Processing an order. Onboarding a customer. Producing a management report. Approving expenditure. Choose something that matters and that people know reasonably well. Then describe how you believe it currently operates. Do not ask AI to improve it yet. Instead, ask it to question your description. Where does the information come from? Where does it go next? Where does somebody wait? What gets checked?
What gets entered twice? What happens when the standard route fails? Where do people exercise judgement? Which activities appear to compensate for problems somewhere else? Then take those questions back into the organisation. Speak to the people doing the work. Follow a few real examples. Look at documents and data. Correct your assumptions. You will probably discover that the process is both simpler and more complicated than you thought. Simpler because some activities may serve little useful purpose. More complicated because experienced people are doing things that never appear in the formal description.
Only then ask what should change. Perhaps AI should perform part of the work. Perhaps conventional automation is enough. Perhaps two systems need connecting. Perhaps information needs capturing differently. Perhaps an approval should disappear. You may discover that the best intervention contains no AI at all. That is entirely acceptable. The purpose of using AI on the business is not to create more AI projects. It is to improve our understanding of how the organisation works so that we can make better decisions about what to change. You do not need to understand the whole organisation to begin. You need one real piece of work and enough curiosity to ask:
What is actually happening here?