Why Most AI Consultants Get It Wrong (And How to Find the Right One)

Why Most AI Consultants Get It Wrong (And How to Find the Right One)

Why Most AI Consultants Get It Wrong (And How to Find the Right One)

The AI consulting market is crowded right now. Two years ago, these people were selling digital transformation. Before that, they were selling cloud migration. The words change. The dynamic doesn't.

I'm an AI consultant. I should probably not be writing this post. But I've been in IT for 23 years — most of it in manufacturing, where I watched company after company get talked into technology projects that delivered vendor revenue and not much else — and I think business owners deserve a straight answer about what to look for and what to run from.

Why Most AI Consultants Get It Wrong

They start with the technology, not the problem.

The most common mistake I see is consultants who walk in with a preferred platform or toolset and build a solution around it, regardless of whether it fits your actual situation. They're solving for their expertise or their vendor partnerships, not your business.

The tell: they start talking about specific tools before they've spent any real time understanding how your business operates. A good AI engagement starts with 2–4 weeks of listening, not proposing.

They overscope to protect revenue.

There's a real incentive to make AI projects bigger than they need to be. More complexity means longer engagements means more money. A consultant with integrity will tell you when a $15,000 implementation is the right answer, even when they could sell you a $75,000 one.

The tell: the proposal is comprehensive and impressive-looking, but when you ask "what will I specifically have in 90 days?" the answer is vague.

They don't understand your industry.

AI for a law firm looks different than AI for a contract manufacturer. The workflows are different, the data is different, the compliance environment is different, the people are different. A generalist consultant who's never worked in your space will spend the first half of the engagement learning things you already know — and billing you for it.

The tell: they struggle to give you concrete examples of how they've solved similar problems in similar industries.

They underestimate change management.

This is the one that actually kills implementations. You can deploy the most technically sound AI solution in the world and have it collect digital dust because nobody changed how they work.

Getting a team to adopt new tools and new workflows is a human problem, not a technology problem. It requires time, communication, training, and honest conversations about resistance. Most consultants either don't know how to do this or treat it as an afterthought.

The tell: their proposal has no explicit section on training, adoption, or change management. "We'll train your team" buried in the scope is not a plan.

They disappear after go-live.

AI systems need tuning. The first version of anything is rarely the right version. What works in month one often needs adjustment in month three as you see how it performs in the real world.

A consultant who hands you the keys and vanishes is leaving you with a depreciating asset. Good implementations include a defined post-go-live support period and clear documentation so you're not dependent on them forever.

What a Good AI Consultant Actually Does

The best implementations I've been part of had a few things in common:

They started small and proved value early.

Instead of a six-month implementation with a reveal at the end, they delivered something usable within the first 30–60 days. Something small enough to be evaluated and adjusted. Proof of concept before scaling.

They were willing to say no.

Good consultants will tell you when an idea won't work, when a vendor is overselling, or when your business isn't ready for a specific approach yet. If someone agrees with everything you propose, they're not adding value — they're just telling you what you want to hear.

They built internal capability, not dependency.

The goal of good consulting is to leave you better equipped to manage the solution yourself. Not to create a situation where you need to keep paying for ongoing hand-holding. Proper documentation, proper training, and a clear handoff are signs someone is thinking about your long-term success, not just their contract.

They talked to the people doing the work.

The biggest insights in any business process improvement project come from the people actually doing the work — not from management's description of how things work. Good consultants spend time on the floor (or on the call center, or in the service bay) before they design anything.

They were honest about what AI can't do.

AI has real limitations. It's not magic. There are things it does well and things it does poorly, and a straight-talking consultant will tell you both. Be suspicious of anyone who sounds like an AI infomercial.

The Questions to Ask Before Hiring Anyone

Before signing a contract with any AI consultant, get direct answers to these:

1. What specific outcome will I have in 90 days?

Not "a roadmap." Not "a foundation." What tangible thing will exist and what will it do?

2. What does your experience in my industry look like?

Ask for specific examples. Ask for references from similar companies.

3. What does your post-implementation support look like?

What's included, what's not, what does ongoing engagement cost, and what documentation will you leave behind?

4. What happens if it doesn't work?

How do they handle underperformance? What's their process for diagnosing and correcting problems?

5. When was the last time you told a client not to pursue an AI project?

This one is telling. If they can't give you an example, they're either too new or they never turn away business.

Why I Work Differently

I spent 23 years in manufacturing IT before moving into consulting. That background means I approach technology the same way a plant manager approaches equipment: it either solves a problem, or it doesn't belong on the floor.

I don't have vendor relationships that influence my recommendations. I don't sell platforms. I assess your specific situation, recommend what makes sense for your scale and budget, help you implement it properly, and make sure your team can actually use it.

My first step with every new client is a free assessment — no proposal attached, no sales pressure. It's an honest look at where AI can and can't help your business right now. You leave with clarity, regardless of whether we work together.

That's what a real first conversation should look like.

→ Ready to talk to someone who'll tell you the truth? Start with my free AI readiness assessment — a no-pressure conversation that gives you a clear picture of where to go next.

Greg Ganfi is an AI implementation consultant with 23 years of manufacturing IT experience. He helps small and mid-size businesses implement AI solutions that deliver measurable results — without the hype.