AI Agency Red Flags: How Not to Get Ripped Off

The AI gold rush has brought out both brilliant operators and a lot of chancers. Telling them apart before you sign is worth real money, because the wrong agency leaves you dependent, overcharged, or holding a strategy deck that does nothing. Here are the seven red flags that reliably mark an agency to walk away from.

Here are the seven red flags, and the healthy opposite of each. This is the mirror image of the ten questions to ask a partner.

Red flag 1: No clear deliverable

If you cannot get a straight answer to "what exactly will I have at the end," walk away. Vague deliverables mean vague value and open-ended fees. A good agency names the working system you will own, in plain terms, before you sign.

Red flag 2: Refusal to fix scope

An agency that insists everything must stay open-ended is often protecting an open-ended invoice. Budgets balloon where scope is fuzzy. A confident agency will agree what is being built and what it costs upfront, and carry the overrun risk themselves.

Red flag 3: Dependency by design

The subtle, expensive one. Some agencies build so you can never leave: undocumented systems, everything in their accounts, no handover. The retainer never ends because it cannot. Ask "could I run this without you," and if the honest answer is no, that is a leash, not a partnership.

Red flag 4: They do not use agents themselves

An agency selling AI that does not run AI in its own business is guessing. Ask what agents power their own operation. If they cannot answer, you are paying them to learn on your money rather than applying what already works for them.

Red flag 5: Hand-waving on data and compliance

Your agents will touch personal data, so an agency should talk easily about minimising it, proper provider agreements, EU hosting, and separating client information. If data and compliance get a vague reassurance instead of a clear process, assume it has not been thought through.

Red flag 6: Jargon and hype instead of plain answers

If every answer is buzzwords and grand promises rather than plain explanations of what will be built and why, be cautious. Operators who understand their work can explain it simply. Hype is often used to cover a lack of substance underneath.

Red flag 7: Strategy deck, no working system

Some agencies sell you an impressive AI strategy, a document, a roadmap, a workshop, with no working agents at the end. Strategy has its place, but if the deliverable is paper rather than a system that saves you hours, you have bought a report, not a result.

A tangible example: imagine Declan's near miss

Imagine Declan, close to signing with a slick agency promising to transform his business with AI. He runs through the red flags and notices three: no clear deliverable, refusal to fix the price, and everything to be run through the agency's own accounts. It is easy to picture where that leads, months of fees, a dependency he did not want, and nothing he owns. He walks, finds a partner who fixes scope and hands him a system in his own accounts, and gets a working result for less. The red flags saved him.

What a good agency looks like instead

The healthy opposite of all seven is consistent: a good agency names a clear deliverable, fixes the scope and price, builds so you own and could leave, uses agents in its own business, handles data with a clear process, explains things plainly, and hands you a working system rather than a document. If an agency ticks those, the absence of red flags is itself a strong signal. Judge on what you walk away owning and how comfortable they are with you being independent. Start on the home page if you want to see the healthy version.

Trust your discomfort

One last practical note: if something feels off in the sales process, it usually is. Pressure to sign quickly, evasiveness on ownership, discomfort when you ask to fix the price, these are your instincts reading the red flags before you have named them. A good agency is relaxed under scrutiny because scrutiny confirms their value; a chancer gets defensive because scrutiny threatens the sale. Give yourself permission to walk away from anything that will not answer plain questions plainly, because with AI, as with any service, the way they sell is a fair preview of the way they will deliver.

How chancers slip past good instincts

Plausible-sounding agencies get past sensible owners in predictable ways, so it helps to know the moves. They lead with impressive-sounding results and case studies you cannot verify, use urgency to shortcut your due diligence, and answer hard questions with confident-sounding jargon rather than plain substance. They often price attractively upfront, then let the open-ended scope and the dependency do the earning later. None of this is obvious in a slick sales call, which is precisely why the seven red flags are worth checking deliberately rather than trusting the vibe. Slow the process down, ask the plain questions, and insist on plain answers, and the chancers reveal themselves because their whole approach depends on you not looking too closely.

The cost of choosing wrong

It is worth being clear about what a bad AI agency actually costs you, because it is more than the fees. You lose the money, yes, but you also lose the months spent on a system that does not work or that you cannot run yourself, the opportunity cost of not having automated properly in that time, and often the confidence to try again, since a bad first experience sours many owners on AI entirely. That last cost is the cruellest, because the technology would have worked with the right partner. Checking the red flags before you sign is cheap insurance against an expensive, discouraging detour that can set your business back a year.