Blog / Guides & decisions
Five questions to ask before hiring an AI or automation company
By Leena Gopinath, DeeAutomation · 25 August 2026 · 3 min read
The short answer
Ask five things: how will you understand our process first, where will our data go, what exactly will we receive and own, how will we know it worked, and what happens after launch. Good answers are specific, plain-spoken and willing to admit limits. Vague or overconfident answers are a warning sign.
How to use these questions
Put the same five questions to every supplier you speak to and write down the answers. The comparison will tell you more than any brochure. Listen for specifics, not enthusiasm. A good supplier will be comfortable saying what they do not know yet.
- How will you understand our process before proposing a solution?
- Where will our data go, and who can see it?
- What exactly will we receive, and who owns it?
- How will we know whether it has worked?
- What happens after launch?
1. How will you understand our process first?
A good answer describes a discovery step: talking to the people who do the work, looking at real examples and mapping exceptions before any design. It should include your staff, not just management.
Be cautious if they propose a solution in the first conversation, or if the plan skips straight to tools. The right technology depends on the process, not the other way round.
2. Where will our data go, and who can see it?
A good answer names the systems and any third-party AI services, says where data is processed and stored, explains who has access, and covers retention and deletion. For personal data they should mention UK GDPR, data processing agreements and impact assessments without prompting.
A poor answer is a general assurance that everything is secure. Ask follow-ups until you can repeat the data flow in your own words.
3. What exactly will we receive, and who owns it?
A good answer lists deliverables: working workflows, source code or configuration, documentation and training. It says who owns the work, whether you depend on the supplier’s own accounts, and what you would need to move to another provider.
Watch for fuzzy scope, and for arrangements that leave you unable to maintain or transfer the work.
4. How will we know whether it has worked?
A good answer agrees measures before building: time per item, error rates, turnaround, volumes handled. It includes a baseline and a review point, and it acknowledges that results depend on your data and adoption.
Be wary of guaranteed outcomes. Nobody can promise results before they have seen your process and data.
5. What happens after launch?
A good answer covers monitoring, support hours, how problems are reported and fixed, what is included and what is charged separately, and how changes to connected systems are handled. They should also explain how you can end the arrangement.
If nobody has thought about maintenance, the automation will gradually break and your team will quietly go back to the old way.
Frequently asked questions
Should I ask for references?
Yes, where possible. Speak to someone about communication, how problems were handled and whether results lasted. Be aware that some work is confidential, so a supplier may offer alternative evidence such as a walkthrough.
How many suppliers should I speak to?
Two or three is usually enough to compare approaches without wearing out your team. Give each the same brief and the same questions so the answers can be compared fairly.
Is a fixed price a good sign?
It can be, if the scope is clearly defined and changes are handled openly. A fixed price on vague scope just moves the argument to later. Ask what is included, excluded and how variations work.
Want to talk it through?
Tell us what is slowing your business down. You will get a straight answer on whether automation or AI is the right fix, and what a sensible first step looks like.
Ask us these questionsRelated guides
This article is general information, not legal, financial or tax advice. Check current official guidance for your situation.