By Joann S. Lublin
Ever wish you had a twin to share your workload?
A small but growing number of executives have done just that, creating AI versions of themselves that offer a glimpse of future workplaces where one person's output is no longer limited to one person.
Here is how it works: An AI system analyzes how an executive writes, speaks and thinks, by studying everything from work emails the person has written to his or her speeches and interviews. Then, the AI double takes on various jobs for the executive -- like answering questions from subordinates -- that use the human's knowledge and communication style. Sometimes, with a video-based version, these AI twins even speak at conferences or make presentations.
Reid Hoffman, a LinkedIn co-founder and partner at venture firm Greylock Partners, mostly uses his digital twin, Reid AI, for public appearances and media interviews. The system -- trained on 22 years of Hoffman's books, speeches, podcasts and articles -- has delivered more than 75 addresses and presentations since its 2024 launch. (A human helped prepare them, though.)
Appearances included a Dubai conference where the digital double introduced itself in French, Chinese and Hindi from a big screen in a meeting room. Reid AI speaks 74 languages. ("I only can speak one," Hoffman says.) The AI also answered English-language questions posed live by its co-creator, Ben Relles.
Thanks to his digital doppelgänger, "I am accomplishing so much more that I couldn't have accomplished before," says Hoffman. "It's probably a 50% time savings on the weeks it's deployed," which is typically every other week.
Doubles multiplying?
Advocates and users say the digital twins have the potential to free executives from routine tasks and leave them available for higher-level work. Digital work twins could become the most significant productivity multiplier since the personal computer, concludes a December report by Kevin Oakes, chief strategy officer at the Institute for Corporate Productivity, a human-capital research and advisory firm.
"Your twin could probably handle half your work, freeing you up to be more creative and strategic," Oakes says in an interview.
In 10 years, Hoffman expects, every company with more than 50 employees will have some highly trained virtual counterparts like his assigned to individuals in senior management, business development, sales, customer service and media relations.
Within a decade, Hoffman predicts, "many large companies will have digital twins for rank-and-file employees as well."
But digital twins also face substantial hurdles -- such as getting the workforce at large to accept the idea. The twins also raise tough questions about how we work. One of the biggest: Will companies start to replace workers with digital copies? And then there is the matter of ownership. Does the company get to keep the accumulated knowledge that goes into a digital twin -- or can the human take that expertise along if he or she leaves?
The (sort of) personal touch
For now, some executives say their twins greatly increase the number of people they can help.
"The way my twin responds to you online is exactly how I would personally," says Bala Sathyanarayanan, chief human-resources officer at industrial-packaging company Greif. His twin, which communicates via electronic messaging, has interacted with nearly 3,300 employees since it launched in December.
Sathyanarayanan's AI twin, BalaBot, was trained on a selection of his publicly available materials, including articles, podcasts and conference remarks. "It does not ingest my private email history, confidential HR case files, employee records or internal communications, " he says.
BalaBot users at Greif often are executives and managers who pick Sathyanarayanan's digital brain for guidance about tough issues like coaching underperformers and advancing their own careers.
Dexter Strong, Greif's senior manager for global learning and development, used one of three strategies suggested by the AI twin to handle a team member who kept missing expectations -- a plan involving weekly check-ins and the possibility of reassignment, among other things.
The team member shifted from "just getting her work done to being a leader in her area," Strong says. "It's a 180-degree turnaround."
Brian Hartzer, chief executive of Quantium Health, says he created his twin, Virtual Brian, in 2024 to "basically act as me." The tool was trained on his leadership book, among other things, and his digital counterpart interviewed him so the executive could refine his own views about various leadership topics.
Alix Duncan, an executive manager at Quantium Health, got Virtual Brian to help her prepare for her recent performance review. (Quantium Health is a unit of Quantium Group, a data-analytics and AI company.) The twin told her that her self-evaluation underrated her accomplishments, because she played down her pivotal role in drafting two successful client proposals.
After interacting electronically with Virtual Brian, Duncan raised her self-assessment's "nice" rating to "great" for that category and added examples of how she leads co-workers, among other things.
Duncan says she felt more confident and less stressed when she spoke to the real Hartzer for her formal review, leading to "probably the easiest performance review I've ever had."
Bugs, not features
Yet AI twins have their limits and may need human oversight to make sure they work properly in key situations. An AI double with a supervisory role could create legal headaches by misstating employment rules, for instance. "That is why digital twins should support human decision-making, not replace legal review or human accountability," says Paul Jurcys, senior intellectual-property counsel at Vinted, an online secondhand retailer.
What's more, a twin may have trouble dealing with tricky communication -- even if it reflects the core logic of how a person decides and acts, says Ashutosh Garg, CEO and co-founder of hiring platform Eightfold AI and a spun-off startup, Viven. Reid AI, for instance, occasionally looks serious after delivering a joke, Hoffman says.
Then there is the classic AI problem of hallucination: Reid AI sometimes makes things up. Or its reply is flat-out wrong on an opinion query like, "What's your favorite ice cream?" (It went with "vanilla," Hoffman says, though "if I had to pick one, it's mint chip. Lack of data here causes it to be off, but it's improving.")
Basic glitches are also an issue. In spring 2025, when Kelly Monahan was managing director of Upwork's research institute, the work marketplace generated digital humans for her and a colleague. Her AI clone, which she dubbed Digital Kelly after quitting Upwork, was supposed to greet about 200 hotel-industry executives at a December conference where Monahan spoke about AI and the future of work.
But Digital Kelly "kept stuttering, repeating the same line," she says. "There was an audible gasp in the room."
Monahan immediately shut down her digital double. "This is why we keep humans in the loop," she jokingly told attendees. "AI isn't ready for the main stage."
Rejecting the copy
Meanwhile, not all humans like dealing with digital twins. Nearly everyone who was offered the chance to have Reid AI as a speaker initially requested Hoffman, a spokeswoman says. In the end, though, they came around -- because, for instance, "they can get Reid AI in locations and times that don't work for me," Hoffman says. (He always speaks at his own annual tech-industry summit.)
Josh Bersin, head of a research and advisory firm bearing his name, initially encountered resistance over his companywide introduction of an AI twin. The product from Viven uses individual employees' personal expertise, emails and other authorized work documents to customize a large language model that can, say, prepare customer email with its human equivalent's typical word choice and priorities.
"Everybody freaked out and said, 'I don't want all my emails in there,' " Bersin says. He assured worried colleagues that they could shield nonwork messages -- such as medical-test results -- from their virtual substitute. They also can mark sensitive work documents as private.
Skepticism nevertheless persists among the roughly 50 staffers. Certain employees use their digital twin fewer than five hours a week on average, estimates Charles Kinzie, an associate director involved with client support. By contrast, Kinzie says, his twin aids him at least a few hours every weekday with client-meeting preparation, trend analyses and project updates.
And digital twins raise even larger issues about the nature of employment. For instance: Will workers be able to take their AI counterparts when they leave a job? And can their former employer keep using their expertise in digital form?
In the future, "I expect to see situations where departing employees will be compensated for leaving their digital twin and associated data and knowledge behind," says Vinted's Jurcys.
Monahan anticipated this issue at Upwork. She helped write a policy that lets anyone with an Upwork virtual counterpart take its likeness and personal expertise when leaving the company. But the company gets to keep the twin's proprietary knowledge about the business.
Since Monahan left in November to become an independent future-of-work researcher, Upwork no longer uses any form of Monahan's digital double, a spokesperson says. But former co-workers can continue using research she did for Upwork. Monahan has retrained her virtual substitute about her forthcoming book and other professional knowledge.
And, of course, twins pose a basic concern that affects all AI: What if businesses deliberately create AI twins to replace employees? "Doing so without clear consent and governance will very likely result in backlash, " says Tori Paulman, an analyst and future-of-work specialist at research firm Gartner.
(MORE TO FOLLOW) Dow Jones Newswires
May 25, 2026 12:16 ET (16:16 GMT)
Copyright (c) 2026 Dow Jones & Company, Inc.
Comments