If you've searched for "AI automation company Canada" or "AI consultant near me" in the last year, you've noticed the field got crowded fast. Every dev shop, marketing agency, and freelancer with an OpenAI API key now calls itself an "AI automation partner."
Some of them are genuinely good. Many are a two-week bootcamp and a landing page.
Here's a practical checklist for telling them apart — written for the Canadian SMB owner who has thirty minutes and no interest in becoming an AI expert themselves.
1. Do they publish real pricing, or just "contact us"?
Opaque pricing usually means one of two things: the vendor prices by how desperate you sound on the call, or they genuinely don't know what things cost because every project is bespoke chaos.
Look for published price ranges at the package level — audit, build, ongoing support. You don't need exact numbers before a call, but you should be able to find a number without booking one.
2. Is there a human being attached to the company?
A consultancy with no named founder, no team page, and no LinkedIn presence is either brand new or deliberately anonymous. Neither is disqualifying on its own, but you should be able to answer: who, specifically, is accountable if this goes wrong?
Look for a named founder or lead engineer with a verifiable background — prior employers, certifications, a LinkedIn profile that existed before last month.
3. Do they know what PIPEDA is without being asked?
If you're a Canadian business handling customer or employee data, your AI vendor should bring up data residency and PIPEDA compliance before you ask about it. If a vendor's answer to "where does our data go?" is a shrug or a vague "it's secure," that's a real problem, not a minor one.
Ask specifically: which cloud region does this run in, and is that documented anywhere?
4. Do they lead with your outcome, or their tech stack?
Watch the language on their homepage and in the first call. "We use cutting-edge LLMs and RAG architecture" tells you nothing about your business. "We cut your reporting time from four hours to ten minutes" tells you what changes for you.
A partner worth hiring can explain what they do without a single acronym, and can also go deep on the acronyms the moment you ask.
5. Is there a real, contained first step — or is the first meeting a sales pitch for a six-figure engagement?
The most common failure mode in this industry is the six-month, six-figure "AI transformation" proposal built on a single discovery call. Reasonable vendors start smaller: a scoped, fixed-price audit or diagnostic that tells you whether it's worth building anything at all — including telling you "no" when that's the honest answer.
If every conversation ends in a large proposal with no off-ramp, that's a sales process, not a diagnosis.
6. Can they show you something that actually shipped?
Demos are easy. Production systems handling real data, real users, and real edge cases are not. Ask specifically: what's the oldest system you've had running in production, and what's broken since launch? A vendor with real deployments will have an honest answer to the second half of that question. One with only demos and pilots usually won't.
7. Do they tell you when not to automate something?
This is the simplest tell in the whole list. A vendor whose answer to every problem is "yes, we can build that" is optimizing for the sale, not for you. Somewhere in a real audit or discovery process, a good partner will tell you a process isn't worth automating, or that a cheaper fix exists than the one you asked about.
A Quick Checklist
- Published pricing at the package level (not just "contact us")
- A named, verifiable founder or lead engineer
- Raises data residency / privacy compliance unprompted
- Leads with outcomes, not acronyms
- Offers a small, fixed-price first step before any large proposal
- Can point to a system that's been in production for a while
- Will tell you when something isn't worth automating
None of these individually disqualifies a vendor. But a company that's missing most of them is optimizing for closing the deal, not for your outcome.
CODIA is a Canadian company — founder-led, Canadian data residency by default, fixed-price packages published on every service page. Book a free intro call and judge us against this same checklist.