A demo is easy to build and tells you very little. The first meeting tells you more, because you can see what the person across the table is curious about.
A good consultant starts with your business. They want to know what is costing you the most time or money right now, and they want it in specifics: which task, how often, who does it. If they open with what their tool can do, they've picked the answer before hearing the question.
Next they'll ask how the work gets done today, step by step. You'll end up describing the workarounds, the spreadsheet someone keeps on their own laptop, the message that always gets forwarded to the wrong person. They need all of that before they change anything, because the workarounds are usually there for a reason.
Then comes success. Ask what it would look like and how you'd both know. Fewer missed enquiries, a few hours back each week, fewer mistakes on invoices. It should be something you can check a month after launch.
The question we'd push hardest on is what happens when the AI gets something wrong, because it will, sooner or later. Where does the customer's conversation go? Who finds out? How does a person take over, and how quickly can the problem be resolved? A vague answer here is a warning.
You should also hear one about your customers' information: where it is stored, which services process it, and what happens to it if you stop working together. Our post on POPIA covers why that matters.
And a good consultant will sometimes ask whether you need AI at all. A better process, a shared spreadsheet or an automatic reminder can do the job for a fraction of the cost. A consultant who never suggests that is worth a second look.
You should expect us to answer all of this too.