There is no shortage of firms offering AI and automation work in Australia right now. Very few of them will tell you what they are not good at, which makes the buying decision harder than it should be.

These are the questions we would ask if we were on the other side of the table.

On scope

What is the first deliverable, and is it a document or a working system. Which process are we starting with and why that one. What is explicitly out of scope for stage one.

On ownership

Whose accounts hold the workflows. Where does the code live. If we part ways in month four, what do we keep and what stops working. Who on our side is named as owner, and what is expected of them each week.

On data

Where does our data go, which providers process it, and under what terms. Is any of it retained for training. What is the answer we can give a customer who asks whether their information touched a model.

On failure

What happens when the system cannot complete a task. Who is notified, how fast, and how is it cleared. What is the expected exception rate in the first month, and how will we know if it is worse than promised.

On after launch

Who fixes it when a supplier changes a format. Is support a retainer, an hourly arrangement, or goodwill. What does month six look like when the novelty is gone and it simply has to keep running.

The signal to watch for

A partner worth having will narrow your scope in the first conversation rather than widen it. Enthusiasm for everything you mention is not confidence, it is a lack of experience with what these projects cost to finish.