It may sound like something from a science fiction novel, but ‘digital doppelgangers’ are on the rise. Business insights company Gartner listed digitally replicating employees as one of its Future of Work Trends for 2026.
As we witness the emergence of digital ‘actors’ and ‘musicians’ into the mainstream, Gartner reported that digital doppelgangers are appearing in the broader workforce too. Researchers suggest that organisations update their AI governance to protect and reward employees’ likeness, as AI is increasingly shaped in their image.
Although the concept of digital doppelgangers is not new – LinkedIn co-founder Reid Hoffman posted a YouTube video of himself in a Q&A with his AI twin back in April 2024 – evolutions in the technology mean that its potential in the workplace is growing.
The available technology to create a ‘digital double’ is advancing beyond merely replicating what someone produces, such as their notes or emails, to incorporating more realistic behaviour and tone too. “We’ve started to see technologies advance and become more accessible,” says Kaelyn Lowmaster, research director in Gartner’s HR practice. “We’ve also started to see more and more organisations [using this technology] in different areas,” including HR leaders creating their own doppelgangers, Lowmaster adds. The opportunity
Peterborough City Council hit the headlines last summer when it turned a knowledgeable staff member, Geraldine Jinks, into an AI chatbot. Having worked 35 years at the council, there was an opportunity to transfer and retain her knowledge and skills using an AI-powered knowledge assistant designed to reflect Geraldine’s spirit and helpfulness.
Geraldine’s case highlights the knowledge management and skill retention angle to adopting this technology. Organisations are increasingly exploring the use of digital twins and AI avatars to replicate high-performing employees.
“It’s the classic situation we’ve had in HR for many years: how do we transfer knowledge from this high-performing, long-serving employee that we think we can’t do without?” highlights Nic Elliott, HR director and head of employment at law firm Actons Solicitors. The same question applies to roles that are ‘one deep’, where a single person is a main source of knowledge, such as a business owner whose expertise and legacy could be retained by creating a doppelganger.
With general fears around a growing over-reliance on AI and skill atrophy, creating digital doppelgangers of professional experts to mentor colleagues is perhaps an appealing proposition. “You could get their knowledge and expertise without overburdening the experts that you have in the organisation,” highlights Lowmaster.
This is an increasingly attractive prospect as we head into an ageing population, with experts on the brink of retirement and organisations identifying a skills precipice.
Adam Dustagheer, managing director of data and AI consultancy Datnexa, says that his company – which partnered with Peterborough Council to deliver Hey Geraldine – has had interest from organisations across sectors in using this technology to address the ‘grey wave’.
There’s value in having a knowledge base that can help futureproof teams. “A lot of people have to know a lot for their jobs these days, and you can’t retain all of that information,” reflects Dustagheer. Doppelgangers could have all of that knowledge on tap.
Double vision, double output?
Digital twins could also be seen as a heightened productivity driver, particularly by those at executive level. Sam Liang, CEO of transcript software company Otter.ai, built an avatar of himself to help handle routine meetings. The CEOs of both payment provider Klarna and video conferencing tech firm Zoom have also used their digital doubles to make announcements.
There are massive efficiency savings to be made by using digital twins, suggests David Clarke, director at language and software company Guildhawk, which provides a ‘digital humans’ service. Thanks to multilingual AI models, Guildhawk clients use the tech to deliver presentations in multiple languages, Clarke explains. “Imagine trying to do that in the old way; it’d take you weeks to record them all,” says Clarke.
Last year, Swiss bank UBS started deploying cloned versions of its analysts to share with clients, citing time and efficiency savings along with an increased client demand for video format as reasons for its rollout. However Jon Dawson, CPO at international hospitality company Lore Group, challenges whether the adoption of AI and avatars does actually give people back time, or merely create more work.
The conclusions from HR leaders at an AI summit he recently attended were that, although AI use sped up outputs, this caused people to work “faster, harder and actually spend less time thinking,” reflects Dawson. This seems to counter the benefits AI is said to bring.
Creating a clone
Digital doppelgangers can take different forms. Hey Geraldine is an AI chatbot whose digital twinning element involves gathering a person’s data – such as through written or recorded interactions – which can be fed into a large language model that can then pick up that person’s knowledge, as well as their patterns of behaviour and likeness.
An AI avatar and digital twin can then be combined to create a fairly realistic online version of that person. There have even been cases of HR leaders worried they had interviewed a candidate’s digital doppelganger during an interview process, says Lowmaster.
Companies like Delphi offer an AI-powered digital cloning service, while in the case of UBS, analysts visited a studio where AI video creator Synthesia made avatars of them, capturing their voice and likeness. Then a language model was used to read analysts’ reports and generate a script to create a realistic AI-generated video. HeyGen is another AI video generator that provides a similar service.
Many businesses also build digital twins with their internal IT engineering teams. Before hiring consultancies or software providers, Lowmaster recommends assessing exactly what it is an organisation wants to use digital twins for, and working with IT teams to see if existing tools might be able to deliver those goals.
Synthesia’s senior strategic advisor Kevin Alster believes the next decade will be defined by a shift from “static, one-way content to interactive, conversational experiences powered by AI agents”. The company is advancing its own offer further, to create a more “responsive and contextual” experience, Alster reports.
Lay the groundwork
Datnexa worked with Peterborough City Council to build a knowledge base for Geraldine, drawing from her first-hand insights, local policy and practice. A key initial consideration was to make sure that it was a tool that staff would actually use, highlights Dustagheer, explaining that they asked themselves: how do we create something that can be trusted by staff?
Context was also considered in initial stages of creating the knowledge base. “It’s not just as simple as dumping a load of content into something, it requires a contextual layer,” he says, which is something we’re starting to see more of among AI tooling.
Building in context helps users answer questions that require the contextual experience that someone like Geraldine could input around job-specific examples. Working with team members in the follow-up months for weekly feedback huddles was also crucial, as it built trust and confidence, which then drove adoption.
Additionally, establishing strong ownership and senior oversight, as well as a clear brief on what problem the technology sought to solve was, is important, states Dustagheer. He adds: “Define what is involved, and bring people with you.”
Early testing of Hey Geraldine found savings of 15 minutes per conversation for occupational therapists. Results suggested that responses were consistent with (real) Geraldine’s advice, which also helped boost trust.
Key considerations
Data protection was a top consideration for HR professionals using this technology. The data set used to train a digital twin is “gold dust”, stresses Clarke. It’s why you won’t see many people’s AI avatars publicly; they will be kept within an internal firewall. He recommends that other companies wanting to create and use digital twins to do the same.
“Lock your data down,” he advises, and make sure it’s “verified, validated and tested”. Nothing should change in that vault “until you are 100% happy that the answer is going to be correct,” he says.
Identity fraud is a top concern for Catrin Gaston-Penny, HR director at biotechnology company Cell and Gene Therapy Catapult. This issue makes her “uncomfortable” about the idea of twinning individuals.
The use of deepfake bots is already a threat for organisations. In 2024, CEO of advertising group WPP, Mark Read, was targeted by an elaborate deepfake scam involving AI voice cloning in an unsuccessful fraud attempt. After the attack, he warned employees about techniques that “go beyond emails to take advantage of virtual meetings, AI and deepfakes”.
To strengthen identity verification mechanisms, Gaston-Penny suggests working closely with an organisation’s IT department and a cybersecurity team. Additionally she recommends employers looking into this technology to start off with transparent data collection policies, robust privacy controls and proper ethical AI review processes.
Ensure you’ve got policies in place, and deal with the topic with sensitivity, she suggests. “If you’re going to be utilising an individual and their skill set, then be really cognitive of how that is being done,” she adds.
Built-in bias is also something to consider, and an issue that has already arisen in the use of AI in recruitment. As humans have inherent biases, it’s possible this could be transferred to a digital doppelganger, reflects Elliott, which poses a discrimination risk for HR. Employers should therefore be mindful of equality and discrimination legislation.
Organisations should also establish who will have human oversight for the tool, says Gemma O’Connor, head of HR advisory and technical services at HR software firm BrightHR. They could, for example, appoint a data protection officer to ensure that policies are consistently being reviewed and that the content remains up-to-date.
And what about hallucinations – when AI produces fabricated outputs? Clarke assures that the risk of this can be reduced to near zero by having a grounded training model. Although this may not remove the anxiety of wondering what exactly your doppelganger is saying on your behalf.
Questions to address
There seems to be a myriad of questions for HR leaders to chew over when thinking about implementing this technology with an employee. HR experts stress locking down an initial employee agreement that addresses the details around issues like compensation, how an employee will get paid for the training of their digital twin, as well as what happens post-employment.
If you’re going to use a worker’s image, employers have got to ask: Do you have the right documentation in place? Do you have their explicit permission? Have you factored in what happens if they’re to leave the company? says O’Connor.
She questions: “If my face was being used, and I was to then leave the company, what happens? Do they continue to use my image or my employee doppelganger?”
Employers should be mindful of who they choose to twin in the first place, and manage the people issue side of it, she highlights. Employees will want to know whether the technology is going to support their role, or result in them taking on extra duties.
This begs the question of why an employee would want to be replicated in the first place. “The obvious risk is you won’t be needed anymore,” notes Elliott, who ominously compares it to the outsourcing of staff seen today, such as when UK workers are sent to train their replacements in another country where the company can get cheaper labour. The rollout of digital twins could lead to employee suspicion about what the motivation behind it is.
In the case of Hey Geraldine, the tool had a precisely targeted job: “It’s not a Swiss Army knife, it’s a scalpel,” highlights Dustagheer. “It’s very much focused on that one job.”
Having a clear understanding of why the tool is being introduced can help address concerns. It can also be seen by some employees as providing a chance to create a professional legacy.
Human touch
“We shouldn’t just look at this as a technology to replace people. It’s a tool to augment people,” stresses speaker and business transformation expert Allister Frost, who uses a lowkey digital doppelganger of himself to assist with tasks by feeding his work into a generative AI assistant.
But when it comes to rolling out digital twins at an organisational level he warns: “You need to really ensure that as the technology gets smarter, the humans don’t get smaller.”
He wouldn’t, for example, send a digital doppelganger to deliver a talk on his behalf. “Absolutely not, because that’s not me,” Frost insists.
“The human touch is still important,” agrees Gaston-Penny. “People don’t have to be everywhere all the time, but showing up in person helps.”
Dawson has come across similar technology used in recruitment to interview on someone’s behalf. This can look and speak like a person, but it makes him wonder what a candidate would make of this: “[If you sent your clone to an interview,] would you feel that the organisation really values you as an individual?”
Is the future cloned?
Conversations have already taken place at Dawson’s workplace about the potential of this technology from a customer service perspective. But the technology wasn’t developed enough at the time, to answer the breadth of questions they required. However, with the current rate of AI development, Dawson thinks “it’s probably something we’ll revisit again.”
In a role that requires frequent international travel, wouldn’t Dawson himself benefit from a digital twin? He’s on the fence: “I would like to see it being successful and well implemented, and see the results, before we went down that path.”
He’s mindful of how the technology could evolve going forward, and presents a long-term thinking approach: “Is it worthwhile spending all the time and resources creating something like that when actually in three years’ time it could significantly change again?”
Considering its potential in an HR role, O’Connor is unsure whether a digital doppelganger could handle the broad range of queries when there’s not always a black and white answer or linear process. “I don’t know how AI would adapt in those situations,” she worries.
However she can see its potential for dealing with specific tasks such as customer complaints or FAQs. “There’s potentially a space there for it, once you have the allocated resources for the human oversight element.”
Elliott acknowledges the benefits of digital doppelgangers but reflects that it feels like “quite a minefield to navigate well”. With Gartner putting digital twins just at the beginning of their ascent on the Gartner Hype Cycle, which tracks the maturity of emerging technology, its potential impact on the workplace remains in the hands of the future.
Legal implications
California is one of the only places that has laws related to digital twins. Designed to protect actors and artists by giving them more control over their digital likeness, the state’s Assembly Bill 2602 established legal requirements for employers and their employees regarding the use of a worker’s ‘digital replica’.
Equity, a trade union for creative workers, has called on the UK government to introduce a new system of automatic rights over people’s voices, faces and body. These ‘personality rights’ would allow everyone to have control over their identity in the age of AI. In the meantime, there are other pieces of legislation to consider.
Gemma O’Connor, head of HR advisory and software firm BrightHR, highlights the Data Use and Access Act 2025, which contains updated laws around digital information in relation to GDPR and privacy. “Whilst it will give a bit more flexibility when it comes to automated decision making,” she says, a key piece of the legislation is that it strictly requires human oversight and putting appropriate safeguarding measures in place, “so you still need human intervention.”
Additionally, if a company has operations in the EU, there’s the EU AI Act to consider, which highlights “quite heavy penalties” for data protection breaches. As data protection legislation is different to employment law, O’Connor stresses the importance of HR professionals working with other teams in the business, whether its legal, risk or compliance, to confidently approach the topic.