Most Businesses Are Using AI Wrong

Almost every company I talk to has AI now. Somebody has a ChatGPT account, somebody else pays for Claude, and the founder has asked it to write a few emails. Then nothing much happens. The tool sits there like an expensive intern who was never told what the company does.

That is not a hunch. An MIT study published in 2025 found that 95 percent of corporate generative-AI pilots produced no measurable financial return. Boston Consulting Group surveyed more than 1,250 companies and found that only 5 percent were getting value from AI at scale, while 60 percent reported little or none. McKinsey found that 88 percent of organizations use AI somewhere, but only about 6 percent see it make a real difference to profit.

Those are big companies with budgets, consultants and whole teams assigned to the problem. Small businesses are not doing better. In a 2026 survey of small-business owners, 70 percent said they needed more training to use AI effectively. Most of them were learning from YouTube and social media, and a third were asking ChatGPT how to use ChatGPT.

The problem isn't the AI

The researchers keep arriving at the same conclusion. The MIT authors blamed a "learning gap" in how organizations use the tools, not the quality of the models. Gartner went further and predicted that through 2026, companies will abandon 60 percent of AI projects that aren't supported by AI-ready data.

AI-ready data is a dull phrase for a simple idea. Your AI can only be as smart as what it can see. Out of the box, it sees nothing about your business. It doesn't know your products, your clients, your test results, your vendors, your prices or the decision you made last Tuesday and the reason you made it. All of that is scattered across shared drives, inboxes, old proposals, voice memos and the heads of two or three people who are tired of being asked.

So people do the only thing they can. They paste a document into the chat window, get a decent answer, and start over the next day. That isn't a system. It's a habit, and it doesn't compound.

What I see from the inside

I have worked in and alongside operations and digital marketing departments for 25 years, usually as the person who ran the whole web and digital operation, which meant I was also the person everyone asked when they couldn't find something. I know what scattered knowledge costs because I spent years being the human search engine.

Recently I worked with a marketing agency owner who sells digital services for a living. He had never heard of the tool his own competitors were using to build websites in a day, and he didn't know that Claude can connect directly to a company's files. He is a smart operator. He simply had no foundation, so every new tool looked like a toy.

A founder's company records were spread across two computers, two cloud services and a USB drive. I merged about 2,000 files, roughly 50 GB, into one organized knowledge base, checked that nothing was left behind, and set up his own Claude to work from it with standing instructions.

The ladder

I think about this as a ladder with four rungs.

On the first rung, a company is chatting. People use a generic AI to write and summarize, and it knows nothing about the business.

On the second rung, a company is feeding. People paste files in by hand. The answers improve, but they are only as good as whatever someone remembered to paste, and nothing carries over.

On the third rung, a company is grounded. Its knowledge lives in one organized place, and its own AI reads it directly. Answers come from the company's own files, with sources attached. This is what I call a Business Knowledge Base.

On the fourth rung, a company is running. The team works from the knowledge base every day, and automations built on top of it handle routine work.

Nearly every company I meet is on the first or second rung. The studies suggest that is where most of the failed projects come from. Companies are buying more powerful engines and never building the road.

Where to start

The fix isn't a bigger AI budget. It starts with three plain questions. Where does your company's knowledge actually live today? Which of it would you need if the person who knows it quit tomorrow? Can your AI see any of it?

If the honest answers are "everywhere," "most of it" and "no," you're in good company, because that's where most companies are. It also means the biggest gain available to you right now has nothing to do with which model you choose. It comes from organizing what you already know so that a machine can read it.

That is the work I do. If you want to know which rung your company is on, I built a two-minute scorecard that will tell you.

Sources: MIT NANDA, "The GenAI Divide," 2025. BCG, "The Widening AI Value Gap," October 2025. McKinsey, "The State of AI," November 2025. Thryv, "2026 AI and Small Business Adoption Survey," April 2026. Gartner press release, "Lack of AI-Ready Data Puts AI Projects at Risk," February 26, 2025.