20 Questions to Ask Before Hiring an AI Professional
Interviewing AI professionals can feel intimidating if you're not technical. But the right questions cut through jargon and surface whether someone has actually shipped working AI systems versus run a few demos. These 20 questions are organized into four buckets — experience, approach, technology, and pricing — with brief guidance on what a confident, credible answer sounds like in each section. No technical background required.
About Their Experience (Questions 1–5)
1. What industries have you implemented AI solutions for?
2. Can you share a case study similar to our project — same industry or same problem shape?
3. What was the measurable outcome of your last AI project? (look for numbers, not adjectives)
4. How many AI implementations have you completed end-to-end versus advised on?
5. What's the biggest AI project that didn't go as planned, and what did you learn?
What to listen for: specific industries with relevant constraints, named outcomes ("cut processing time from 4 days to 6 hours"), and a candid failure story. Anyone who claims they've never had a project go sideways is either green or selling you something.
About Their Approach (Questions 6–10)
6. How would you approach our specific problem?
7. What data would you need from us to get started?
8. How do you handle situations where the data isn't ready?
9. What does your typical project timeline look like, and where does it usually slip?
10. How do you ensure knowledge transfer so we're not dependent on you forever?
What to listen for: a methodology, not a magic answer. Strong consultants describe a discovery phase before committing to architecture, ask about data quality before pitching tools, and have a documented handoff plan. Watch for anyone who jumps straight to recommending a specific platform — that's a red flag they're selling tools rather than solutions.
About the Technology (Questions 11–15)
11. What AI tools or platforms do you recommend, and why those for our case?
12. Will we own the solution, or is it proprietary to you?
13. How do you handle data privacy and security, especially for regulated data?
14. What happens when the underlying model needs to be retrained or updated — including frontier-model upgrades from providers like Anthropic, OpenAI, and Google?
15. How does the solution integrate with our existing systems (CRM, ERP, ticketing, data warehouse)?
What to listen for: model-agnostic thinking. Good consultants in 2026 don't lock you into one model vendor — they architect so you can swap Claude, GPT, or Gemini as pricing and capability evolve. They should also have a clean answer on data ownership and IP.
About Pricing and Terms (Questions 16–20)
16. What is your pricing model — hourly, project-based, retainer, or outcome-based?
17. What's included in the price, and what costs extra (model API fees, infrastructure, additional integrations)?
18. What are the ongoing costs after the initial implementation goes live?
19. What does your contract look like — exit terms, IP ownership, and what happens to the system if we part ways?
20. Do you offer a paid pilot or proof-of-concept phase before full commitment?
What to listen for: clear separation between fixed-scope work and ongoing operating costs (model API spend especially). The strongest consultants offer a small paid POC because it protects both sides — you reduce risk; they get paid for discovery rather than sinking it into a free pitch deck.
How to Conduct Reference Calls
Always ask for references — and actually call them. Aim for 2–3 references from projects similar in scope to yours. Ask each reference: How did the consultant handle scope changes? Was the timeline realistic and did they hit milestones? How was their communication during the hard parts of the project? What's still working well, six or twelve months in? Would you hire them again — and if not, what would you change?
The most useful question is the last one. "Would you hire them again?" forces a binary judgment that's harder to fudge than open-ended praise. If a reference hesitates, asks for more time, or gives a soft answer, treat that the same as a no.
Frequently Asked Questions
Frequently Asked Questions
That's a red flag. A good AI professional should be able to explain their experience, approach, and pricing in plain language. If they hide behind jargon, get defensive, or can't give a straight answer about pricing or IP ownership, move on — there are plenty of qualified professionals who will.
Yes. Ask for 2–3 references from projects similar to yours in size and complexity. Specifically ask references about communication, deliverable quality, timeline adherence, and whether they'd hire the consultant again. References that have only good things to say but no specifics are often coached — push for stories.
Three is the right number for most projects — enough to compare approaches and pricing, few enough that the evaluation doesn't drag for months. For larger engagements (>$100k), expand to 4–5. For very small projects (<$15k), getting three quotes can cost more in your time than the project savings; one or two well-vetted candidates is usually fine.