Hiring an AI consultant is a leap of faith in a way hiring an accountant usually isn't. Most business owners have a rough sense of what good accounting looks like. Far fewer have a feel for what a well-run AI project looks like, which makes it easy to end up mid-engagement with a partner who's long on buzzwords and short on delivery. The good news: the warning signs tend to show up early, before you've spent real money — if you know what to look for.
They Promise a Timeline Before They Understand the Problem
If a consultant hands you a fixed price and delivery date in the first conversation, before asking a single question about your data, your systems, or what "done" means to you, treat that as a red flag rather than a sign of confidence. Good AI work almost always starts with discovery — understanding what you have to work with and where the real complexity is hiding — because that's often where a project's true cost and timeline live. A partner who skips straight to a number is either guessing or padding the estimate to cover the guess. Either way, you're the one who absorbs the surprise later.
Every Answer Is "Yes, We Can Build That"
Ambition is good. Reflexive agreement is not. A consultant who never pushes back, never asks whether a feature is really necessary for version one, and never says "that's harder than it sounds" is either inexperienced or telling you what you want to hear. The partners worth working with will occasionally talk you out of something — not because they can't build it, but because they can see it's not where your budget or attention should go yet. If every conversation is frictionless agreement, ask more pointed questions and see what happens.
They Can't Explain Their Approach in Plain Language
You don't need to understand the internals of a language model to hire someone to build with one, but a consultant worth trusting should be able to explain, in language you understand, what they're building and why it's the right tool for your problem. If every explanation collapses into jargon the moment you ask a follow-up question, that's often a sign the underlying thinking is thinner than the vocabulary suggests.
There's No Plan for What Happens After Launch
A launched AI system isn't a finished one. Models drift, data changes, and edge cases surface once real users start interacting with something instead of a demo audience. If a proposal covers building and delivery but goes quiet on maintenance, monitoring, or iteration afterward, you're being sold a launch, not a working system. Ask directly: who watches this after it ships, and what does that cost? A vague or dismissive answer is worth taking seriously.
They Want to Skip the Discovery Phase Entirely
Related to the first point, but distinct: some consultants actively resist a discovery or scoping phase because it eats into billable build time or delays a signed contract. That instinct is backwards. A short discovery phase almost always pays for itself by catching data problems, unrealistic scope, or a mismatched solution before anyone's committed to a full build. A partner who treats discovery as a nuisance rather than the foundation of a good project is optimizing for closing the deal, not for your project succeeding.
Pricing Is Opaque, or Suspiciously Simple
Real AI projects have real cost drivers — data cleanup, integration work, testing, iteration — and those vary project to project. Be cautious of a proposal that reduces everything to one flat number with no breakdown, and equally cautious of one so vague it can expand indefinitely without anyone noticing. You should be able to see roughly where the money goes.
They Can't Point to Anything Concrete About How They Work
You don't need slick case studies to trust a partner — plenty of good consultants are early in building their track record. What you should expect is a clear, concrete description of their actual process: how they scope work, how they communicate during a build, how they handle scope changes, what a typical engagement looks like from kickoff to handoff. Vagueness about process, as opposed to vagueness about outcomes, is the harder red flag to excuse.
Trust Your Own Discomfort
None of these signs are disqualifying on their own — a rushed answer in one meeting doesn't mean a partner is wrong for you. But if you notice more than one of these patterns, or find yourself quietly uneasy without being able to name why, that instinct is worth listening to. The cost of walking away before signing is small. The cost of discovering the mismatch three months into a build is not.
If you're evaluating AI partners right now and want a second opinion on a proposal, or just want to talk through what a well-scoped engagement should look like, feel free to book a short call or reach out through our contact page. No pressure, no pitch required.