AI Digital Twins: The Billion-Dollar Category Nobody Has Claimed Yet

01 /The Concept

Every organization hemorrhages knowledge – every time an employee is sick, on vacation, or simply in the wrong time zone, and every time someone leaves. Viven's answer is a digital twin: an AI version of each employee that can be queried in their absence.

The immediate use case covers availability gaps – the twin answers colleagues' questions when the employee is unreachable. The longer-term play is what Viven calls "knowledge that never walks out the door": even after an employee leaves, their twin stays queryable on unfinished projects, workflows, and institutional context.

For incoming employees, the digital twin of their predecessor accelerates onboarding dramatically. Instead of piecing together institutional knowledge from scattered documents, a new hire can simply ask.

The twin connects to the employee's digital workspace – email, document storage, meeting recordings, and the applications they use. In doing so it doesn't just access the same information the employee has; it learns to think, write, and communicate in that person's style.

The twin is also useful in day-to-day work, not only in emergencies. Employees can use it as an extra pair of hands – to remember something, draft a message, prepare for an upcoming meeting, or execute an action in a tool.

Viven was founded by the creators of Eightfold, which was valued at $2.1 billion on its last funding round in 2021. They originally built it as an internal tool for their own company and spun it out as a separate startup earlier this year. The founders currently split their time between the two companies, with Eightfold formally listed as one of Viven's first clients.

Other clients have since started coming on board, and Viven made its public debut alongside the announcement of a $35 million seed round from several well-known venture funds.

02 /Why It Matters

Here's how the founders frame the problem Viven addresses:

- "Knowledge leakage" – the inability to retrieve expertise from colleagues when needed – costs companies $31.5 billion per year.

- Employees at global organizations work across time zones, which routinely blocks access to information that's needed to keep processes moving.

- Corporate knowledge platforms collect "averaged" company-wide information, while individual employee expertise is often unique, difficult to extract, and harder still to transfer.

One customer – a director at a company with a market cap of $6.79 billion – put it plainly: "This platform gives us the speed and flexibility we need."

The obvious objection is that a digital twin isn't a real employee, so its answers might be incomplete, wrong, or occasionally hallucinatory.

But that calls to mind a passage from Guy Kawasaki's old book "The Macintosh Way" about the corporate culture that made Apple work. How do you rank the options "right answer," "wrong answer," and "no answer"?

Kawasaki's view: a wrong answer beats no answer. Any answer contains some information and leads to action. No answer is zero information and an excuse to stall.

In practice, if an employee needs a colleague's input to move forward but can't get it, the work stops – because there's no answer to check off. A wrong answer, by contrast, removes that excuse. And real employees give wrong answers too, so the bar isn't perfection.

Digital twin platforms are emerging, but many are targeting the creator economy – offering influencers the ability to "clone" themselves so the twin handles fan Q&As and student queries. Delphi ([related review](/review/staryj-infobiznes-umrjot-no-rynok-to-ostanetsja)) built that and raised $16 million in new funding in June.

Some startups that started there are now pivoting toward the enterprise. Amigo ([related review](/review/ii-experty-eto-sovsem-ne-ii-sotrudniki)) raised its first $6.3 million in November 2024 on a Delphi-like platform for experts and consultants wanting to scale their reach with digital twins – and has since pivoted to selling that platform to clinics, where it powers digital twins of doctors, nurses, and other patient-facing staff.

03 /Opportunities

A digital twin is fundamentally different from an AI assistant. A twin preserves the unique opinions and skills of a specific person. An assistant aggregates averaged, generalized knowledge.

For questions like "what's two plus two," the distinction doesn't matter. But there's an enormous range of situations where it matters a great deal.

Humy ([related review](/review/tvoego-rebjonka-mozhet-uchit-aleksandr-makedonskij)) built a platform with digital twins of historical figures – Alexander the Great, Winston Churchill, and others – that can be deployed in schools, universities, and museums to share first-person perspectives on events and engage in debate.

Remark ([related review](/review/kak-nanjat-togo-kogo-nanjat-ne-mozhesh)) raised $16 million in July on a platform with expert digital twins acting as online consultants for e-commerce – offering product recommendations based on the personal experience of the expert who served as the twin's model. One of those experts is a member of the US Olympic team.

Hyperbound ([related review](/review/fishka-nuzhna-chtoby-zacepitsja)) raised $15 million in September on a sales training platform where reps practice conversations with AI characters – rude, polite, dismissive, methodical, and everything in between. The platform can also ingest a customer's messages and meeting transcripts to create a digital twin of that specific client, helping reps prepare for the real conversation.

Despite all of this, digital twins remain a significantly underexplored category compared to AI assistants. Every month brings dozens of new AI assistants; genuine digital twins are rare. The applications clearly exist, and they extend well beyond the examples above.

What domain would you start building AI twins for? And if you know the answer – what's stopping you from building the platform?