A major new study from learning company Section is flashing warning lights at every business leader who thinks buying AI tools is the same as deploying AI capability.

The 2026 AI Proficiency Report, published July 7, found that while 69% of workers say their company has taken some action on AI agents, fewer than 10% of those same workers can correctly define what an AI agent actually is. Just 3.8% can write instructions to build a simple AI automation.

These numbers matter because companies are not waiting for readiness. They are buying agent platforms, building workflows, and measuring productivity gains, all while the people those systems are supposed to help are largely flying blind.

The Proficiency Gap in Numbers

Section assessed workers across hundreds of enterprise organizations and sorted them into four categories: Novices, Experimenters, Practitioners, and Experts. The results are a wake-up call.

Only 5.5% of employees reached Practitioner or Expert level, meaning they use AI regularly in ways likely to generate real business value. The vast majority, 73.5%, are what Section calls Experimenters: people who dabble with AI for one-off tasks but have no repeatable workflow. Another 20.9% are Novices who barely engage with AI at all.

Put bluntly: for most organizations, the people who need to deliver AI ROI are not equipped to do so.

Deployment Without Training Is the Expensive Mistake

The report found a clear performance gap between organizations that train their teams and those that do not. Workers at companies that deployed agents and ran training programs scored an average of 47.5 out of 100 on Section’s proficiency benchmark. Workers at companies that deployed agents without training scored 33.1 out of 100.

That 14-point gap is not a number to dismiss. It represents the difference between an AI investment that compounds and one that flatlines after the launch announcement.

The access disparity is also stark. C-suite executives are more than twice as likely as individual contributors to have access to agent-capable AI tools, and five times more likely to have received agentic training. So the people making decisions about AI are the ones who understand it, while the people expected to execute with AI tools are left to figure things out on their own.

Why This Pattern Keeps Repeating

This is not a new problem. Every major technology wave, from ERP systems in the 1990s to cloud adoption in the 2010s, has produced the same dynamic: enterprises buy the tools before they build the skills. The difference with AI agents is that the capability gap is more visible and more expensive.

An ERP system you cannot use sits idle. An AI agent you cannot direct can actively produce wrong outputs, hallucinate workflows, or automate the wrong thing at scale. The risk of undertrained AI use is not just wasted spend. It is operational.

What This Means for Business

If you are a business leader, this report tells you three things.

First, having AI tools is not the same as having AI capability. The tool is the infrastructure; your people are the engine. Without trained operators, you are paying for capacity you cannot access.

Second, the ROI case for AI training is now measurable. The Section data shows a 43% performance improvement between trained and untrained workers in AI-deploying organizations. That is a return most other training investments cannot match.

Third, the gap between the C-suite and the front line is a business risk. When executives understand AI agents and employees do not, you get top-down mandates met with bottom-up confusion. That is the recipe for low adoption, shadow IT, and expensive rollbacks.

The organizations that will win the next three years of AI competition are not the ones that bought the most tools. They are the ones that built the deepest bench of people who know how to use them.

At Enterprise DNA, this is the exact problem we have spent years preparing for. Our platform has helped 220,000 data professionals build the kind of foundational skills, data literacy, structured thinking, and workflow design that make AI agents actually work. Whether you are starting from zero or trying to close the gap between your C-suite and the rest of your team, that foundation is what makes the difference.

The Section report is a snapshot. The question for every business leader reading it is: what are you going to do before the next one comes out?

---

The AI Proficiency Report was published by Section on July 7, 2026. The full report is available at sectionai.com.