Why AI delivers no results (and how to fix it)

You tried ChatGPT. Maybe you even had a chatbot built or bought an AI tool for something. And after a few months? Little to nothing changed. If AI is delivering no results in your business, you are far from alone. In fact, it is the rule rather than the exception. In 2025, research by MIT showed that 95 percent of corporate AI projects deliver no measurable impact on profit. In this article you will read why AI so often delivers nothing, and, more importantly, how to be in the 5 percent that actually benefits.

95 percent delivers nothing, and it is not the technology

The figure comes from an MIT report with a telling title: The GenAI Divide (2025). The researchers spoke with 150 business leaders, surveyed 350 employees and analysed 300 AI projects. The finding was blunt: about 5 percent of projects produced clear revenue growth, the rest stalled with no measurable effect. And the cause? Not the quality of the AI. According to the researchers the problem is the learning gap: the tools do not adapt to the way your company works, and the company does not learn how to use the tool well.

Analyst firm Gartner sees the same pattern. In June 2025 it predicted that more than 40 percent of so-called agentic AI projects will be scrapped before the end of 2027, due to rising costs, unclear business value and inadequate control over risk. In short: it is not that AI does not work. It is that most companies use AI the wrong way. And the good news: those mistakes are avoidable.

Reason 1: you started with the tool, not the problem

This is the mistake that causes most of the others. A company hears that AI is the future, buys a chatbot or a tool, and only then starts wondering which problem it actually solves. Gartner says it plainly: unclear business value is one of the main reasons projects fail. You started with the answer instead of the question.

The lead researcher of the MIT report put it well about the companies that did succeed: they pick one pain point, execute it well, and partner smartly with a party that knows their work. One pain point. Not 'AI across the whole business', but that one process that currently costs you the most time or money.

Reason 2: ChatGPT is great for you, but does not know your business

Many owners think: my team already uses ChatGPT, right? True, and for one-off tasks it is excellent, drafting an email, summarising a text, translating something. But that very flexibility is the weakness the moment you want to run a business process with it. The MIT report points exactly at this: generic tools like ChatGPT stall in companies because they remember nothing of your way of working. Every time you start over. There is no connection to your systems, no knowledge of your customers, no fixed route.

A tool that does deliver results is embedded in how you work: it knows your quote template, knows which system an order should land in, and follows the steps you normally do by hand. That is the difference between a handy aid for one employee and an automation that genuinely takes work off your plate. Read more in our article about ChatGPT for your business.

Reason 3: you put AI where there is least to gain

This is where whole countries go wrong. According to Statistics Netherlands (CBS), Dutch companies use AI most often for marketing and sales (36.4 percent), followed by administration and management (30.3 percent). Sounds logical, sales is visible and appealing. But the MIT report found that the biggest gain is not in marketing, but in the back of your business: administration, quotes, order processing, the work you now outsource or do by hand.

In other words: most companies put their AI budget precisely in the corner where there is least to gain, and leave the boring processes, where the real hours sit, untouched. That is a waste. The customer service that runs into the evening, the quotes left lying around, the data retyped from one system into another: that is where your time savings are.

Reason 4: you build it yourself instead of having it built

One of the most striking findings in the MIT report: companies that buy AI from a specialised party and partner with them succeed about 67 percent of the time. Companies that try to build it themselves succeed only a third as often. Yet many organisations choose the do-it-yourself route to save costs, with an employee who 'knows a bit about it' putting in their evenings.

Reason 5: nobody owns the result

An AI project that belongs to nobody dies. The MIT report found that successful companies placed ownership with the people on the floor, the team lead who works with the process daily, instead of with a central 'AI department' far from practice. And almost nobody measures beforehand how much time or money a process costs now. Without that baseline you cannot see afterwards whether anything improved. Then the feeling lingers that AI delivers nothing, while you simply never measured it.

How to make AI actually deliver

The 5 percent who make it work do a few things differently. None of them is technically complicated.

• Start with your biggest time drain, not with a tool. Which work costs the most hours now and does not really belong to you or your team?
• Pick one process. Solve that well before you start the next. One pain point, well executed.
• Measure first. How many hours a week? How many missed enquiries? Put it on paper so you can see the difference later.
• Make sure it is embedded. The solution should know your existing systems and follow the steps you now do by hand.
• Have it built by someone who understands your business. Buying and partnering works twice as often as doing it yourself.
• Keep it up. A process changes, your customers change. Without maintenance every automation sags over time.

What a failed AI project really costs you

An AI project that delivers nothing has not failed for free. You keep paying the monthly subscriptions, you sank your team's hours into it, and, perhaps most costly, you now feel that AI does not work for us. That last part stops you from trying a second time, properly. Meanwhile the competitor who does it right is winning hours every week on the same work you are still doing by hand.

Start with an honest review

I start with a free introductory conversation. We discuss what takes time, which systems you use and where automation could help. You then receive a proposal with automation opportunities, integrations, a schedule and costs.

• 1. Getting acquainted: Where does your time go, and which systems do you use? We discuss how your business works.
• 2. Opportunities & proposal: You receive a proposal covering automation opportunities, connections, timing and costs.
• 3. Building & testing: I build your system step by step and keep you updated. We test, improve and expand it.
• 4. Live & beyond: Once everything runs well, we launch the system. You track your AIs and their actions in your own dashboard.