← Back to blog

The One AI Advantage Your Competitors Can't Buy

John Abbitt

·

·

9 min read

Every business now has access to the same powerful AI. The research shows the lasting advantage belongs to companies whose people know how to use it, check it, and build it into how work gets done

The question I hear more than any other from business owners is some version of "Which AI tool should we be using?" It's a reasonable question, and I understand why people ask it. There are hundreds of products on the market, new models come out every few weeks, and nobody wants to bet on the wrong one.

My answer usually surprises people. I'm happy to tell you which tools we like, and I'll tell you for free. Our team will write about them openly on this blog and on social media. The reason we can be so relaxed about it is that the tool is the part of AI that gives you the least lasting advantage. Your competitors can buy the same one tomorrow, often for less than you paid. The thing that separates the businesses getting real results from AI from everyone else is their people: whether the team knows when to use it, how to check it, and how to build it into the way work actually gets done.

That's the argument I want to make in this post, and I want to back it up with more than my own opinion, because I run an AI transformation firm and you should expect me to say people matter. So let's look at what the research actually shows.

What a moat is, and why AI makes a poor one

Warren Buffett popularized the idea of an economic "moat." The picture is a castle with a wide trench of water around it. A moat is whatever protects a business from competitors, something they can't easily copy. It might be a brand that customers trust, a location nobody else can get, relationships built over decades, or a way of doing the work that took years to perfect.

Now apply that test to AI. It's hard to think of a technology in history that has become cheaper and more widely available faster. Epoch AI, an independent research group that tracks the economics of artificial intelligence, found in 2025 that the price of reaching a fixed level of AI performance was falling somewhere between 9 and 900 times per year, depending on the task. The same model that felt cutting edge last spring is a budget option this fall. And the very best tools on the market are available to a two-person shop and a Fortune 500 company on roughly the same terms.

When everyone can buy the same thing at the same price, owning it stops being an advantage. It becomes part of the cost of doing business, like having a phone system or a website.

We've seen this argument before

This idea has a history. In May 2003, a writer named Nicholas Carr published an essay in Harvard Business Review with a deliberately provocative title, "IT Doesn't Matter." His argument was that as computers, software, and networks became cheaper and more standardized, they stopped giving any one company an edge. Once every business could buy the same technology, he said, it behaved more like electricity or railroads: essential to have, and nearly impossible to win on.

The essay caused an uproar. Executives at Microsoft, Intel, and Hewlett-Packard pushed back hard, and plenty of critics argued Carr had gone too far. I think both sides had a point, and the way they were both right is what matters for business owners today. Carr was correct that simply owning the technology stopped being an advantage. His critics were correct that some companies still pulled far ahead with it. The companies that pulled ahead did so because of how they used the same technology everyone else had: how they redesigned their work around it, and how well their people understood it.

AI is following the same path, only much faster. Owning it will matter less every year. Using it well will matter more.

Where the advantage actually shows up

If the tool itself is a commodity, you'd expect the research to show big differences in results among companies using the same tools, with those differences traced back to people. That's exactly what it shows.

One of the most important studies on AI at work was done by economists Erik Brynjolfsson, Danielle Li, and Lindsey Raymond, and published last year in the Quarterly Journal of Economics. They studied more than 5,000 customer support agents at a company that rolled out an AI assistant. On average, the agents resolved 14 percent more issues per hour. For newer and less experienced agents, the improvement was 34 percent. The most interesting part is why. The researchers found evidence that the AI was essentially passing along what the company's top performers knew to everyone else. In their words, the tool "disseminates the best practices of more able workers." The value in that system came from people. The AI was the delivery mechanism for knowledge that experienced employees had built up over years.

A second study shows the flip side. In 2023, researchers from Harvard Business School and several other universities worked with Boston Consulting Group on an experiment with 758 of its consultants. On tasks that AI was well suited for, consultants using it completed 12.2 percent more work, finished 25.1 percent faster, and produced results rated more than 40 percent higher in quality. On a task the researchers deliberately chose because it fell just outside what AI could do well, the consultants using AI were 19 percentage points less likely to get the right answer than those working without it. The researchers called this the "jagged technological frontier." AI is brilliant at some things and confidently wrong at others, and the line between the two isn't obvious. The only thing that protects a business from the confidently wrong part is a person who knows where that line is.

Gartner's research on workers points the same direction. Its 2026 survey of more than 12,000 employees and managers across 40 countries found that employees who are proficient with AI across multiple uses are twice as likely to be highly productive, 2.3 times as likely to deliver high-quality work, and 3.2 times as likely to drive effective process improvements. Employees with a positive outlook toward AI were 3.4 times more likely to be highly productive. Gartner also found that 73 percent of the most productive AI users are managers or executives, which tells me the skills are pooling at the top of most organizations instead of reaching the people doing the daily work.

Put those three findings together and a clear picture emerges. The same AI produces very different results depending on who's using it, what they know, and how they feel about it. That difference is the moat.

What it costs when the people part gets skipped

The research on what happens without that human foundation is just as consistent.

Researchers at Stanford's Social Media Lab and BetterUp coined a term last year that has stuck: "workslop." It describes AI-generated work that looks polished but lacks real substance, the kind of report or email that seems finished until someone has to actually use it. In their survey of 1,150 full-time U.S. workers, 40 percent said they had received workslop in the past month, and each instance took close to two hours to sort out. The researchers estimated the cost at about $186 per employee per month, or roughly $9 million a year for a company of 10,000 people. Nearly half of the people who received it said they saw the sender as less creative and less reliable afterward. Every one of those hours traces back to people using a powerful tool without the judgment or guidance to use it well, and no choice of model would have prevented it.

IBM released a large study of its own just last week, based on surveys of 1,500 chief human resources officers and 8,800 employees around the world. Seventy-one percent of the HR leaders said the most essential skill in an AI-enabled workplace is the ability to supervise, validate, and override what AI produces. Only 29 percent of employees ranked judgment as important. Forty-three percent of employees said the blame falls on them when an AI system fails, and 80 percent of HR leaders acknowledged that AI is creating "invisible" work, meaning all the checking, correcting, and context-giving that nobody budgets for. The gap between what leaders know matters and what employees have been prepared to do is exactly where AI investments leak value.

Even the AI companies are selling people

If you want to know what the industry really believes, watch who it hires. This summer, TechCrunch reported that forward-deployed engineers, meaning technical experts who work directly inside customer businesses to make the technology fit, have become the AI industry's hottest hire. OpenAI and Anthropic have both built out teams for this kind of work. According to a study cited in that report, the share of companies planning to hire for the role jumped from roughly 5 to 10 percent at the start of the year to 70 percent by the end of the second quarter.

Think about what that means. The companies that build the most advanced AI in the world, with the most to gain from convincing you the product is all you need, are hiring people to sit with customers and help them use it. They've learned what every successful implementation eventually learns: the model gets a business to the starting line, and people carry it the rest of the way.

How to tell whether you're building a people moat

"Invest in your people" is advice everyone agrees with and very few companies act on, so I want to make it concrete. Here are five questions I'd ask about any business trying to get real value from AI. They work whether you have ten employees or a thousand.

Can your people tell when the AI is wrong? Given what the Harvard and BCG study found, this is the first thing I'd check. Someone on each team should know the kinds of mistakes AI makes in your specific work, whether that's a misread invoice, a made-up part number, or a confident answer to a customer question it should have passed to a person. That knowledge comes from practice on real examples, and it's worth more than any feature on a vendor's product sheet.

Is it safe to push back on it? IBM's study found that when HR leaders share responsibility for AI decisions, 76 percent of employees feel safe questioning AI recommendations, compared with 43 percent when HR only plays an advisory role. People who are afraid of looking like they're resisting change will let bad output slide. A team that's encouraged to challenge the tool will catch problems early.

Have the skills spread past the enthusiasts and the executives? Most companies have a few people who love AI and figure it out on their own. That's a great start, but a fragile one. If Gartner is right that the most productive users are concentrated in management, the biggest untapped value in most businesses sits with the front-line people who haven't been shown how the tools fit their jobs.

Is your best people's know-how being written down? The customer support study is a reminder that AI is at its most powerful when it spreads what your top performers already know. That only happens if someone takes the time to capture how your best dispatcher, estimator, or bookkeeper actually thinks, and turns it into shared instructions the whole team and the tools can use.

Would it survive switching tools? This is my favorite test. If you had to replace your AI vendor next quarter, would your team's skills, habits, and documented processes come with you? If the answer is yes, you've built something that belongs to you. If everything would have to start over, what you have is a subscription.

Where Avolis stands

We built Avolis around a belief that I expect will keep us somewhat out of step with the market for a while. A lot of the industry is racing to package AI transformation into a product you can buy and install. We think the part that makes it work, the training, the coaching, the follow-up three months in when the excitement has worn off, is the part that deserves the most attention.

Here's what that looks like in practice. We start by finding where the work in your business actually gets stuck. We choose the tools that fit, and we explain why. Then we build the workflows, integrations, and agents that connect those tools to the way your team really operates. We train each role on its own work, write down what your best people know so the whole company can use it, and set clear rules for what AI handles on its own and what needs a person's approval. And we stay accountable for the results until the new way of working is simply how your company runs. Choosing the tools is the smallest piece of that list, and it's the only piece a competitor can copy.

So ask us what tools we recommend, and we'll tell you. Then, when you're ready to build the part your competitors can't copy, a good place to start is our AI Readiness Evaluation. It gives you an honest look at where your team stands today and what it would take to bring them along.

Sources

The One AI Advantage Your Competitors Can't BuySep 29, 2026
The Service-as-Software Model Has a People ProblemSep 24, 2026
The Next Customer for Your Software Is an AI AgentSep 23, 2026
Your Next Coworker Is an AI Agent. Here's How to Onboard It Well.Sep 16, 2026
What actually happens in an AI Readiness EvaluationJun 18, 2026
Forward-deployed, explained without the jargonJun 4, 2026
The first workflow you should automate, and the ones you should notMay 21, 2026
Why your last AI pilot stalled after the demoMay 7, 2026
Buy, build, or embed: how to bring AI into your businessApr 23, 2026
Map the workflow before you automate itApr 9, 2026

Ready to make AI work for you?

Book an AI readiness evaluation. If there’s nothing worth automating, we’ll tell you.

Start the conversation