AI Power Users Are Rapidly Outpacing Their Peers. Here’s What They’re Doing Differently

Last November, consulting firm EY surveyed 15,000 employees across 29 countries about how they use AI at work. The results should worry every founder: 88 percent of workers now use AI tools daily, but only 5 percent qualify as “advanced users” who’ve learned to extract real value. That 5 percent? They’re gaining an extra day and a half of productivity every single week. The other 95 percent are stuck using AI for basic search and document summarization, essentially treating a Ferrari like a golf cart. When OpenAI released its State of Enterprise AI report in December, it confirmed the same pattern. Frontier workers—those in the 95th percentile—send six times more prompts to AI tools like ChatGPT than their median colleagues. For coding tasks, that number explodes by 17x. If these AI tools are identical and access is universal, why are the results so wildly different for workers around the world? And what separates power users from everyone else? Ofer Klein, CEO of Reco, a SaaS security platform that discovers and secures AI, apps, and agents across enterprise organizations, offers some insights into what sets the power users apart. 1. They experiment while others dabble High performers treat AI tools like junior colleagues they’re training. They iterate on prompts rather than giving up after one mediocre response. They’ve moved beyond one-off queries to building reusable prompt libraries and workflows. The rest of your team tried AI once or twice, got underwhelming results, and concluded it wasn’t worth the effort. What they don’t realize, however, is that AI requires iteration. The first response is rarely the best response. Power users ask follow-up questions, refine their prompts, and teach the AI their preferences over time. 2. They match tools to tasks Power users typically maintain what Klein calls a “barbell strategy”—deep mastery of one or two primary tools plus five to eight specialized AI applications they rotate through depending on the task. “They’re not trying every new AI that launches, but they’re not dogmatically loyal to one platform either,” Klein explains. “They’ve developed intuition about which AI is best for what.” They might use ChatGPT for brainstorming, Claude for analysis, and Midjourney for visuals. Most employees, by contrast, force one tool to handle everything. When it inevitably underperforms on tasks it wasn’t designed for, they blame AI rather than their approach. 3. They think about work differently It’s easy to assume that the biggest behavioral difference between these power users and frontier workers is their technical skill. But, interestingly, it’s not. Rather, it’s how power users think about tasks. They break projects into discrete steps: research, outline, first draft, and refinement. Then they deploy AI strategically at each stage. Instead of asking AI to “write a report,” they ask it to summarize research, suggest an outline, draft specific sections, then refine tone. They understand where AI adds value and where human judgment matters. “The highest performers spend more time on strategic work because AI handles the grunt work,” Klein says. “They use AI to augment their expertise, not replace thinking.” The hidden cost Why does all of this matter? Here’s the math that should worry you: OpenAI’s data shows workers using AI effectively save 40-60 minutes daily. In a 100-person company where 60 employees barely touch AI, you’re losing 40-60 hours of productivity every single day. Over a year, that’s 10,000+ hours—equivalent to five full-time employees’ worth of work you’re paying for but not getting. Meanwhile, your competitors’ power users are compounding that advantage daily. What you can do about it Klein recommends tracking time saved, not just usage frequency. Someone using AI 50 times daily for spell-checking differs fundamentally from someone using it five times to restructure a client proposal. In addition, run an “AI show and tell” where employees demonstrate one workflow where AI saves them meaningful time. You’ll quickly identify who’s truly leveraging these tools versus who’s dabbling. Then, create small cross-functional “AI councils” of five to six employees who meet monthly to share workflows. That should cascade into proper training of employees on how to use these tools the right way. “Only one-third of employees say they have been properly trained,” a BCG survey found. That’s an opportunity forward-thinking leaders can tap into. But don’t just replicate tools; replicate mindset. Giving everyone ChatGPT Plus doesn’t close the gap. The differentiator is teaching people to think in terms of “what can I delegate to AI?” rather than “what can AI do?” The uncomfortable truth, according to BCG’s survey, is that this gap is widest among front-line employees. While more than three-quarters of leaders and managers use AI several times a week, adoption among front-line workers has stalled at just 51 percent. That’s not just a productivity problem. It’s a competitive threat that compounds every quarter you ignore it. Your 5 percent are already working like they have an extra team member. The question is whether you’ll help the other 95 percent catch up before your competitors do. BY KOLAWOLE ADEBAYO, COLUMNIST