Red flags when hiring an AI consultant

What are the biggest red flags in an AI consulting sales conversation?

A polished demo with no baseline

You get an impressive live walkthrough of what the system will do, but nobody mentions the process it's replacing or what that process costs you in time and money today.

Ask: "What's the current baseline, and how did you measure it?" Without a straight answer, there's no way to know afterwards whether anything improved.

"Transformation program" pricing before anyone understands the problem

The proposal is for a multi-month, multi-stage engagement, and it's priced before anyone has watched the actual work happen.

Ask: "What specific process does phase one measure, and what does it cost if I stop after that phase?"

Proprietary middleware and lock-in

The system only runs through the consultant's own platform, so leaving means rebuilding from scratch.

Ask: "If I never speak to you again after this ships, what do I own, and can someone else maintain it?"

Retainers for systems that should run themselves

There's an ongoing monthly fee for a tool that, once built and handed over, has no obvious reason to need paid attention.

Ask: "What exactly does this retainer pay for, month to month?" The retainer itself isn't the problem. A vague answer is.

Jargon as a smokescreen

The conversation is full of terms like "agentic workflows" and "AI-native transformation," and when you ask for a plain-English version, it turns out to describe something fairly ordinary.

Ask: "Can you explain that in one sentence without using the term?" If they can't, they may not fully understand it either.

Predictions that never make it into writing

You hear confident claims about time or cost savings in the meeting, but those numbers don't appear in the proposal.

Ask: "Can you put that number in the contract as the target we measure against?"

Never saying AI is the wrong tool

Every problem you describe gets an AI-shaped answer, whether it fits or not.

Ask: "Can you give me an example of a project you talked a client out of?" A consultant who has never turned down work, or can't name a time they did, isn't being selective.

Are there green flags too?

Yes, and they're mostly the reverse of the red flags. A good consultant wants to watch the actual work before proposing anything. They quote a fixed fee for a defined, measurable piece of work rather than an open-ended engagement, and they hand over the documentation and access you'd need to run the system without them. When you ask about experience, they give you a specific number from a past project instead of a general claim. And they're willing to tell you in writing that an idea isn't worth building, even when that costs them the job.

None of this is unusual. It's what an honest working relationship looks like, written down clearly enough that you can check it.

When is an ongoing retainer legitimate?

Some AI work does need continued involvement. A system built on a third-party model may need updating when the provider changes its API or pricing. A tool that watches live data for anomalies works better when someone adjusts its thresholds as conditions change. And a process that's still changing month to month may need ongoing work rather than a one-off build.

So the question isn't whether there's a retainer. It's whether anyone can explain it. A legitimate retainer is tied to something that will keep changing. An illegitimate one keeps billing for a system that, by any reasonable description, should be running on its own by now. If nobody can tell you what the retainer is paying for this month, you have your answer.

If you're looking at a proposal and something here doesn't add up, send me the brief and I'll give you an honest second opinion.