All posts

Technology

How to spot an AI vendor wasting your money

Not all AI vendors are worth your money. Here are the red flags to watch for, the questions to ask, and how to tell genuine expertise from slick sales pitches.

Mark Blair··7 min read

The AI market is booming, and that means every software company, consultancy, and freelancer has added "AI" to their pitch. Some of them are brilliant. Some of them are selling you a spreadsheet macro with a chatbot stuck on top.

If you are a business owner trying to find the right AI partner, you need to know how to tell the difference. Because the wrong vendor will not just waste your money. They will waste your time, frustrate your team, and put you off AI entirely. And that would be a shame, because the right project with the right partner can genuinely change how your business operates.

Red flag 1: They cannot explain what the AI actually does

Ask a simple question: "Can you explain, in plain English, what the AI does in your product?" If the answer is a wall of jargon about neural networks, machine learning models, and proprietary algorithms, be cautious.

Good AI vendors can explain their technology simply because they understand it deeply. They will say things like: "It reads your invoices, pulls out the key data, and enters it into your accounting system." Not: "Our proprietary NLP engine processes unstructured financial documents through a multi-stage classification pipeline."

If you cannot understand what you are buying, you cannot evaluate whether it is worth the money. And if the vendor cannot explain it clearly, there is a reasonable chance they do not fully understand it either.

The UK government's guidance on buying AI solutions recommends that buyers require clear, non-technical explanations of what any AI system does and how it makes decisions. That advice is aimed at public sector procurement, but it applies equally to private businesses.

Red flag 2: They promise specific results before understanding your business

"We'll save you 40% on admin costs." "You'll see ROI within 3 months." "Our AI reduces errors by 90%."

These might all be true for someone. But if a vendor is making these claims before they have looked at your data, your processes, and your team, they are guessing. Or worse, they are telling you what you want to hear to close the deal.

Honest vendors will say: "Based on what we have seen in similar businesses, you can expect savings in this range, but we need to look at your specific situation before we commit to numbers." That is less exciting, but it is far more trustworthy.

Red flag 3: They want a big contract before proving anything

The AI industry has a term for this: "big bang" implementations. A vendor proposes a six-month, six-figure project to transform your entire operation. It sounds impressive. It is also extremely risky.

Good AI projects start small. A pilot, a proof of concept, a trial on one process or one department. You see whether it works, measure the results, and then decide whether to expand. Any vendor pushing you towards a large upfront commitment without offering a smaller starting point is prioritising their revenue over your interests.

According to Deloitte's global AI survey, businesses that start with focused pilot projects are significantly more likely to achieve positive ROI from AI than those that attempt large-scale deployments from the outset.

Red flag 4: They do not ask about your existing systems

AI does not exist in a vacuum. It needs to connect to your accounting software, your CRM, your email, your ERP, whatever systems you already use. If a vendor is not asking detailed questions about your current technology stack, they are either planning to replace everything (expensive and disruptive) or they have not thought about integration (which means nasty surprises later).

The right question from a vendor sounds like: "What systems do you currently use, and how does data flow between them?" If they are not asking this early in the conversation, they are not planning properly.

It works alongside your existing systems. That is the standard you should hold any vendor to.

Red flag 5: They have no relevant case studies

"We've done this for a Fortune 500 company" is not a relevant case study if you are a 40-person business in the Midlands. Scale matters. Industry context matters. The challenges a multinational faces are fundamentally different from yours.

Ask for case studies from businesses similar to yours in size, sector, and complexity. If they do not have any, that does not necessarily mean they are bad. But it does mean they are less experienced with businesses like yours, and you should price that risk into your decision.

Red flag 6: Everything is proprietary and you own nothing

Ask a simple question: "If we stop working with you, what happens to what you have built for us?"

If the answer involves words like "proprietary platform" and means nothing is transferable, you are buying a dependency, not a solution. You should own the outputs: the integrations, the workflows, the documentation. If the vendor disappears tomorrow, you should be able to hand everything to someone else and carry on.

Good vendors build on open standards and widely supported platforms. They write documentation. They want you to understand what they have built, because they are confident enough that they do not need lock-in to keep your business.

Red flag 7: They disappear after the sale

This catches people out more than anything else. The project goes live, the vendor collects their final payment, and suddenly response times go from hours to weeks. The enthusiastic sales team is nowhere to be found, and you are left with a support email address that nobody monitors.

Before signing anything, ask: "What does ongoing support look like? Who do I contact when something goes wrong? What are the response time commitments?" Get it in writing.

The Federation of Small Businesses digital adoption report found that negative experiences with technology vendors are one of the top reasons UK SMEs delay further technology investments. Getting it right first time matters.

Questions to ask every AI vendor

Here is a checklist you can take into vendor meetings.

About the technology

  • Can you explain what the AI does in plain English?
  • What data does it need, and where does that data come from?
  • How does it connect to our existing systems?
  • What happens when the AI gets something wrong?

About the project

  • Can we start with a small pilot before committing to a larger project?
  • What does the timeline look like?
  • What do you need from our team, and how much of their time will it take?
  • What are the risks, and how do you manage them?

About results

  • What results have you delivered for businesses similar to ours?
  • How do you measure success?
  • When should we expect to see measurable results?

About support and ownership

  • What does ongoing support include?
  • What are your response time commitments?
  • Who owns the system after the project is complete?
  • What are the ongoing costs after the initial project?

What good looks like

A trustworthy AI partner listens more than they talk in the first meeting. They ask detailed questions about your business, your processes, and your goals. They are honest about what AI can and cannot do. They recommend starting small and scaling based on results. They explain things clearly and do not hide behind jargon. They have relevant experience and are happy to share it.

They will also tell you if AI is not the right answer. Not every problem needs AI. Sometimes a simpler process change or a better use of your existing tools is the right solution. A vendor who tells you that, even though it means less work for them, is a vendor worth keeping.

The British Chambers of Commerce procurement resources include guidance on evaluating technology suppliers that is worth reading before you commit to any significant spend.

Want an unbiased starting point?

Our free AI opportunity report gives you a clear, independent assessment of where AI can help your business, before you talk to any vendors. It will help you walk into vendor meetings knowing exactly what you need, what questions to ask, and what results to expect.

Get your free AI opportunity report here

gofasterwith.ai

Mark Blair

Founder, gofasterwith.ai

Frequently asked questions

What is the single best question to ask an AI vendor in a first meeting?

Ask them to explain, in plain English, what the AI actually does in their product. A good answer sounds like: it reads your invoices, pulls out the key data, and enters it into your accounting system. A bad answer is a wall of jargon about neural networks, proprietary algorithms and multi-stage classification pipelines. If a vendor cannot explain it clearly, there is a fair chance they do not fully understand it either, and you cannot judge value in something you cannot describe.

Should I be worried if a vendor proposes a six-month, six-figure project upfront?

Yes. Big-bang implementations are one of the clearest red flags. Good AI projects start with a pilot or a proof of concept on one process or one department, so you can measure results before deciding whether to expand. Deloitte's global AI survey shows businesses that start with focused pilots are significantly more likely to see positive ROI than those attempting large-scale deployments from day one. Any vendor pushing past a smaller starting point is putting their revenue ahead of your risk.

What should we own once the project is finished?

You should own the integrations, the workflows and the documentation, built on open standards and widely supported platforms. Ask the vendor: if we stop working with you tomorrow, what happens to what you have built for us? If the answer involves a proprietary platform that nothing transfers out of, you are buying a dependency, not a solution. A confident vendor does not need lock-in to keep your business and will write documentation that lets another team pick the work up.

How do I check that case studies are actually relevant to my business?

Scale and sector matter more than logo recognition. A Fortune 500 reference is not relevant to a 40-person firm in the Midlands, because the integration challenges, governance overhead and team dynamics are completely different. Ask for case studies from businesses similar to yours in headcount, sector and complexity, ideally with named contacts you can speak to. If they have none, that is not automatically disqualifying, but you should price the added risk into the contract and the pilot scope.

Want to talk about this?

A free 45-minute call to walk through how this applies to your business. No pitch.