Right now there is no higher, or more reassuring, praise than this.

America’s bosses are getting blunt about the reality that AI leads to job cuts. The standard warning goes something like this: If a bot doesn’t replace you, a human who makes better use of AI will.

Cue the race to be seen as a power user, which signifies being one of the humans getting the most out of new technology.

These are the people cranking up productivity, impressing managers and making the rest of us look like slackers. The good news is we can join them.

They aren’t the Ph.D.s who build machine-learning models and command multimillion-dollar pay packages. They are regular workers who have become uncommonly savvy with existing AI tools, often through trial and error.

They get more done, faster, and—this is critical—cultivate reputations for being ahead of the curve.

Krieger, a 27-year-old public-relations representative in San Diego, knows that early-career professionals are among the most vulnerable to automation. She demonstrates her value by sharing AI advice with colleagues.

“I liken it to the way CEOs develop their thought leadership on a topic,” she says. “You’re solidifying yourself as the go-to person for the team on AI.”

Be the real deal

Achieving this status involves personal branding, but you better have the skills to back it up. Sal Abdulla, founder of accounting-software startup NixSheets, discovered many self-styled power users are overhyped when he posted on social media that he wanted to meet some.

He says the ones who impressed him knew how to turn an AI tool into a “brain” connecting several other applications. For example, they might use ChatGPT to automatically import quarterly financial data from Google Sheets into Intuit’s QuickBooks software, then analyze the results against a database of previous earnings reports.

With practice, Abdulla, 38, learned how to do something similar.

“It’s very doable, even though I’m not a software engineer,” he says. “And the opportunistic side of me is realizing that your average accountant won’t be doing this.”

AI power users are always looking for an edge. Kevin Wei is a principal product manager at a software company and runs a YouTube channel called LiftoffPM where he shares tips for landing a job like his. Though he is viewed as a tech authority by his more than 7,000 subscribers, he says he has learned a lot about AI from a friend who studies machine learning in a Ph.D. program.

One big lesson: “It takes practice to know how to write prompts,” says Wei, 29. “AI can do anything you want it to do, but it only knows as much as you tell it.”

A lot of bot-made material is generic, low-quality workslop. Power users get better results by refining their prompts, he says. Instead of simply telling Gemini to compose a message and hoping it matches your vision, he suggests providing specific instructions like the target audience and writing style.

Then it’s time for feedback.

“Give AI reinforcement, like, ‘This was good, but make it shorter’ or ‘Give me three alternatives to choose from,’ ” Wei says.

No instant results

In the time it takes to come up with a detailed prompt and revise the output, you could write the darn message yourself. That’s often how it feels, anyway.

But AI power users are playing the long game.

Aubrey Miller-Schmidt, an insurance marketing manager in Maryland, has been using ChatGPT for a few years. Sometimes she converses with its voice assistant while driving her car.

“I’ll just work through whatever thought process I have going on in my brain,” she says.

Her goal isn’t necessarily an “aha” moment behind the wheel. These conversations help ChatGPT learn about Miller-Schmidt, 44, and respond to her prompts better.

“I have all of this history now with my AI partner,” she says. “It knows how I tend to approach problems and where I tend to get stuck.”

Power users agree it is often more effective to invest in training a primary AI assistant than to play around with many tools that never build big enough sample sizes to understand your preferences.

Jack of all bots, master of none, I guess.

Michael Rueckert’s go-to is Claude, and he stays up until midnight most days “editing, fixing, tinkering and enhancing.”

The late nights are partly because he juggles several jobs as the marketing director of a ski resort in Utah and a hotel chain, plus a one-man startup he launched this year. His business, Centium AI, offers a version of search-engine optimization, helping clients show up in answers to the questions that people pose to AI.

Rueckert, 34, says the productivity boost he gets from Claude is a key to managing his triple duties, but his assistant requires nurturing.

“You’ve got to coach AI and build context to really get it to where you want it,” he says. “Once you put in the time, it feels incredibly magical.”

Hearing power users wax about their long-term relationships with AI assistants is a bit like listening to investors who say the best time to buy a hot stock was years ago.

On the upside, jumping in today puts you ahead of everyone who waits for tomorrow.