"AI consultant" is a new title, and a lot of people have adopted it in the last two years. Some build systems that run inside real businesses. Some run workshops and write strategy decks. Both can be useful, but they're very different things to pay for.
If you're a Calgary business looking for help, here are the eight questions I'd ask anyone you're considering, including me, and what a good answer sounds like.
1. What have you built that's running today?
Not a demo, not a pilot, not a client logo. Ask for something specific that does real work right now, and ask them to show it. A phone agent taking real calls, an invoice system processing real invoices, a booking system a business depends on.
A good answer is specific and a little boring: what it does, who uses it, what broke and how it was fixed.
2. How do you check the AI's work?
This is the most important question on the list. AI makes confident mistakes. Anyone building with it seriously has an answer: automated checks, second reads, comparisons against source documents, human review of flagged items.
If the answer is "the new models are very accurate", keep looking.
3. What does "done" look like?
Ask them to describe the deliverable in terms you could check: this report, produced weekly, in this format, meeting these standards. Vague outcomes like "AI readiness" or "an AI roadmap" are hard to hold anyone to.
4. Where does our data go, and who can see it?
They should be able to say exactly which services will process your data, whether it's used for training, where it's stored and how to delete it. They should also be comfortable working inside your systems when the data can't leave. If they hesitate here, that's a real risk.
5. How do you choose which AI model to use?
The honest answer involves testing on real work and matching models to tasks by quality and cost. Be cautious of anyone loyal to a single vendor, or anyone who says the choice doesn't matter. The models change every few months; the consultant's process for choosing should outlast any one of them.
6. How are you paid?
Hourly billing for AI work can reward slowness, and it puts the risk of overruns on you. Ask whether they'll quote a fixed fee against defined deliverables, and what happens if the result doesn't meet the agreed standard. Someone confident in their delivery should be comfortable with that.
7. What happens after you hand it over?
AI systems need upkeep. Models get updated, your prices change, your software changes. Ask who maintains it, what that costs, and whether your team will understand how it works. You want written procedures and context you own, not a black box.
8. What wouldn't you automate for us?
A good consultant has a clear list of things AI shouldn't do in your business: decisions about people, legal sign-offs, sensitive conversations, anything they can't check. If everything is a candidate for automation, that's a sales pitch, not advice.
A bonus question: will you try it on our real work?
The fastest way to judge a consultant is a small, paid test on one of your actual tasks, with the success criteria agreed up front. A week of real output tells you more than any proposal.
Why local can matter
Plenty of AI work can be done remotely, and much of mine is. But for a first project, being able to sit with the person who does the job today, watch how it's actually done and meet again when you review results is valuable. Most of the rules that make AI work in a business are never written down until someone asks.
How I answer these
For what it's worth, my own answers are on the home page: the voice agent, invoice system, rental platform and studio build line I run, how I test models, and how engagements work, with a fixed fee against agreed deliverables. But use the questions with anyone you talk to. The right consultant for your business will be glad you asked them.