Imagine sending someone who thinks exactly like you to a conference in Dubai, while you are in another meeting. That has already happened. And the one who did it was not a science fiction character, but Reid Hoffman, co-founder of LinkedIn and partner at Greylock Partners.
His digital twin, Reid AI, has given more than 75 presentations since its launch in 2024. It speaks 74 languages. It has answered live questions, has appeared on giant screens in front of executives from around the world, and saves Hoffman, in his own words, about 50% of his time in the weeks it is deployed. "I am accomplishing much more than I could have achieved before," he claims.
The system was trained with 22 years of Hoffman’s books, speeches, podcasts, and articles. It does not improvise: it replicates his logic, his way of communicating, and his positions on the topics on which he has been trained. He built it together with Ben Relles, former head of innovation at YouTube Originals.
Hoffman is not the only one. Bala Sathyanarayanan, head of human resources at Greif, has BalaBot, a twin trained with his public materials that has already interacted with nearly 3,300 employees since December. Managers at the company consult it to resolve complex leadership situations, from how to manage someone who does not meet expectations to how to boost a career within the organization. One documented case: an employee went from meeting the minimum to becoming a reference in his area after his manager applied a strategy suggested by BalaBot.
Brian Hartzer, CEO of Quantium Health, created Virtual Brian to "basically act like me." A collaborator at the company used it to prepare for her performance evaluation, discovered that she was underestimating her own achievements, and arrived at the formal review with more clarity and confidence. "It was probably the easiest performance evaluation I’ve ever had," she said.
But digital twins also fail, and sometimes very visibly. In December 2025, the clone of Kelly Monahan, then the director of Upwork's research institute, was supposed to welcome about 200 executives from the hotel sector. She started stuttering and repeating the same line on a loop. "An audible gasp was heard in the room," Monahan recalls, who shut down the system immediately. Cases like Reid AI also reveal the more well-known limitations of these systems: hallucinations, incorrect responses to opinion questions, and humor that does not always translate well on screen.
Beyond technical failures, twins pose questions that companies still do not know how to answer. Who owns the knowledge accumulated in a twin when the executive leaves the company? Can a company retain the digitized intelligence of someone who no longer works there? Upwork has a clear policy regarding this: the employee takes the image and personal experience, but the company retains the business knowledge. Not all organizations have gone that far in that debate.
A report from the Institute for Corporate Productivity published in December concludes that work digital twins could become the most significant productivity multiplier since the personal computer. Hoffman goes further: in ten years, he expects that every company with more than 50 employees will have highly trained replicas assigned to sales, business development, customer service, and communication profiles. And in the long term, he projects that many large organizations will also have twins for entry-level employees.
From next+, our specialists' analysis points out that the real change is not technological but organizational. Digital twins are not just a productivity tool: they are the first concrete model where an executive's individual knowledge becomes a scalable asset within the company. This transforms the way talent is managed, expertise is transferred, and the impact of a person within an organization is measured. Companies that start building that infrastructure today will have a structural advantage that cannot be bought later: the institutional knowledge of their best profiles, available at scale and without friction.