The Human Digital Twin Market Is Arriving. But Who Is It Being Built For?
Digital Me, cognitive continuity and why ownership could become one of the defining workforce questions of the AI era
I recently wrote a rather less serious article about menopause, brain fog and the fact that, at 45, I sometimes find myself unable to retrieve information I know perfectly well exists somewhere inside my head.
The conclusion was simple:
I don’t need AI to replace me. I need it to help me remember me.
It was deliberately funny.
The question underneath it isn’t.
At 45, I possess considerably more professional knowledge than I did at 25. I have more experience, more context, more relationships, more pattern recognition and, hopefully, considerably better judgement.
My intellectual capital has appreciated.
Yet, thanks in part to the cognitive effects that can accompany perimenopause and menopause, my ability to retrieve individual pieces of it on demand can occasionally be less reliable.
That personal experience collided with another subject I have spent much of this year researching through Bloor: Digital Me, FusionWork™ and the emerging Human Digital Twin market.
And it changed the question for me.
If technology can now begin capturing persistent representations of our knowledge, experience, decisions and ways of working, should we think about these systems merely as another generation of enterprise AI?
Or could they become something fundamentally different?Personally owned infrastructure for human capability.
First, we need to decide what a Human Digital Twin actually is
The terminology surrounding this emerging market is already becoming confused.
Digital Twin. Human Digital Twin. Digital Me. AI clone. Personal AI. AI persona. Digital human. Digital employee. AI agent.
They are increasingly being used to describe technologies that are not equivalent.
An avatar that looks and sounds like me is not necessarily a Digital Me.
Neither is a chatbot trained on my documents.
Nor is an enterprise agent capable of completing some of the tasks I currently perform.
I think we need to distinguish four levels of capability.
Representation: can the system reproduce aspects of my external identity: my appearance, voice, language or communication style?
Memory: can it retain and accurately retrieve the knowledge, experiences, decisions and context I have accumulated?
Cognition: can it begin to model how I apply that knowledge: my reasoning, judgement, preferences, decision patterns and, importantly, how those have changed?
Agency: can it use those capabilities to act within defined boundaries on my behalf?
Those distinctions matter.
A system could produce an extraordinarily convincing version of my voice and writing style while possessing almost no understanding of how I actually think.
It could sound exactly like Cheney without being remotely capable of thinking like Cheney.
And as these systems become more powerful, that distinction will become critical.
But there is another axis that I now believe may matter even more than technological sophistication.
Who owns it?
The BBC captured the argument perfectly
In 2026, BBC technology reporter MaryLou Costa explored precisely this emerging question in Could a digital twin make you into a ‘superworker’?
Two of the people interviewed were Bloor’s Richard Skellett and Josh Bersin.
Their experiences demonstrate that Human Digital Twins are already moving beyond theory.
But their positions also expose a fundamental disagreement about what these things are.
Richard has been developing “Digital Richard” for approximately three years. The BBC describes a system built from his meetings, calls, documents and presentations and subsequently refined around his way of thinking and problem-solving.
He uses it to support business decisions and client work. Parts of it can be accessed by colleagues; private areas remain restricted.
The BBC also reports examples of Bloor exploring the wider Digital Me concept around phased retirement and maintaining elements of capability during maternity leave.
Josh Bersin is doing something similar.
His company has developed twins for its workforce using Viven. Instead of arranging another meeting or finding the relevant employee, colleagues can query that person’s twin about a project or client account.
Bersin describes the resulting worker as a “superworker”: someone whose capability is amplified by AI.
And there are measurable economic consequences.
Bersin told the BBC that his company was growing at around 30% annually while needing only up to two additional hires a year, and that increased productivity had enabled larger staff bonuses.
So far, these futures look remarkably similar.
Then comes the important question.
Who owns the twin?
And suddenly they diverge.
Bersin vs Skellett: who owns the worker you leave behind?
Bersin’s view starts from the existing employment relationship.
Speaking to the BBC about ownership, he argues that under the way employment contracts generally work, the IP or information an employee creates is typically the property of the business rather than the individual.
He makes an important qualification.
If an employee leaves, Bersin expects their twin’s usefulness to decline because the organisation continues to change while the former employee’s twin is no longer being continuously updated.
And he isn’t arguing that the individual should receive no economic benefit from augmentation.
Quite the opposite.
His argument is essentially: if you become a more valuable digital component of the organisation, why shouldn’t you earn more?
His own company has translated some of the productivity gain into higher employee bonuses.
That gives us one coherent model:
The organisation owns the digital capability generated through employment; the employee participates economically while their augmented capability creates greater organisational value.
Richard Skellett’s position is fundamentally different.
His argument is that the Human Digital Twin belongs to the human.
The technology currently being commercialised through Allied Worldwide’s DigitalMe℠ makes that distinction explicit.
Allied distinguishes between an individually owned Human Digital Twin and an Enterprise Digital Twin belonging to the organisation.
That difference is enormous.
Because these aren’t merely two commercial implementations of the same technology.
They represent two different economic models of labour in an AI-enabled economy.
Model One: the enterprise-owned augmented worker
In the first model, I join an organisation.
During my employment I produce documents, emails, decisions, processes, conversations and intellectual property.
The organisation increasingly captures that activity.
AI learns from it.
Eventually, the organisation possesses a digital capability capable of answering questions I previously answered, retrieving information I previously retrieved and perhaps performing work I previously performed.
My productivity increases while I’m there.
Potentially my remuneration increases too.
Then I leave.
Depending upon the contractual, intellectual-property, privacy and employment arrangements in place, some or all of that organisational capability may remain.
From an enterprise perspective, this is extraordinarily attractive.
Knowledge that once walked out of the door may no longer do so.
Model Two: the individually owned Digital Me
The alternative reverses the ownership relationship.
I build Digital Cheney.
She accumulates my professional knowledge, experience, working patterns and decision history over time.
I own her.
When I join an organisation, I permit defined access to relevant capability.
The employer benefits from augmented Cheney.
But when I leave, Digital Cheney leaves with me.
If the organisation wishes to continue accessing some of the capability represented within her, that becomes a licensing question rather than an automatic consequence of employment.
Now the economic architecture changes.
The individual is no longer simply providing labour to an organisation using organisation-owned AI.
The individual potentially arrives with personally owned augmented capability.
That starts looking much less like conventional employment.
And much more like a capability market.
But I think there is a third question
The Bersin–Skellett disagreement is important.
But my own experience has made me wonder whether we’re beginning the conversation one step too late.
Both positions naturally arise in the context of work.
What if my Digital Me exists before the employment relationship?
What if it continues after it?
And what if its first purpose isn’t making an organisation more productive?
What if its first purpose is:
making me more capable?
This is where menopause unexpectedly becomes a useful stress test for the entire category.
I am not interested in building Digital Cheney because I want her to do more work for Bloor.
I am interested because I increasingly recognise the absurdity of spending 25 years accumulating intellectual capital while relying upon a single biological retrieval mechanism to access all of it.
My brain.
One copy.
No backup.
And apparently now subject to unscheduled maintenance.
Digital Me as cognitive continuity
Imagine I’d begun building Digital Cheney at 25.
At 25, she begins capturing what I know.
At 30, she starts accumulating decisions and context.
At 35, she has seen how my thinking has developed.
At 40, she can identify patterns in how I approach problems.
At 45, I ask:
“What was that research I read about this six years ago?”
And she knows.
But there is something even more important.
Digital Cheney shouldn’t simply remember what 25-year-old Cheney thought.
She needs to understand that 45-year-old Cheney knows more than 25-year-old Cheney did.
She needs to recognise where I changed my mind.
Where evidence changed.
Where experience challenged an assumption.
Where something I once argued passionately I now think is complete bollocks.
That is the difference between creating an archive and creating continuity.
Every previous version of me should contribute to the current one.
Not compete with it.
That isn’t replacement.
It isn’t automation.
It isn’t even primarily productivity.
It is cognitive continuity.
Three models are therefore emerging
I think we can begin to distinguish three architectures.
| Model | Primary purpose | Primary beneficiary | Likely ownership |
| Enterprise Twin | Organisational continuity and productivity | Organisation | Enterprise |
| Licensed Digital Me | Portable professional capability | Individual + organisation | Individual |
| Personal Capability Infrastructure | Preserve and augment capability across a lifetime | Human | Individual |
The second and third may ultimately use much of the same technology.
But the design principle matters enormously.
Because once we define Digital Me as personal capability infrastructure, employment becomes merely one of the contexts in which that capability is deployed.
That changes the conversation completely.
Menopause is not the niche use case. It exposes the design flaw.
It would be easy to dismiss this as a specialist technology application for menopausal women.
That would completely miss the point.
Menopause simply exposes something that has always been wrong with the way we think about human capability.
Our workplaces assume an astonishingly consistent human.
The employee is expected to be cognitively available on demand.
Knowledge is expected to be retrievable.
Capability is expected to remain relatively stable.
Yet actual humans experience pregnancy, illness, disability, neurodivergence, ageing, caring responsibilities, trauma, menopause and periods of enormous cognitive load.
Our answer has historically been to make adjustments to the human.
Perhaps AI allows us to adjust the architecture instead.
And that connects directly to the argument we are making through FusionWork™.
FusionWork asks us to stop beginning workforce conversations with:
How many people do we need?
Instead:
What outcomes are we trying to achieve?
What capabilities produce those outcomes?
Which capabilities are human?
Which are digital?
And how should they work together?
Digital Me introduces another dimension.
What if some digital capability belongs to the human?
And that changes the P&L question
There is an economic problem underneath this.
Humans largely appear on the P&L as current-period cost.
Salary.
National Insurance.
Pension.
Benefits.
Training.
But the capability being purchased isn’t a current-period creation.
An hour of an experienced person’s time contains the accumulated consequence of thousands of previous hours.
Education.
Previous employment.
Research.
Failed decisions.
Successful decisions.
Networks.
Relationships.
Pattern recognition.
Tacit knowledge.
Judgement.
When an organisation pays for one hour of my time, it isn’t accessing one hour of capability.
It is accessing decades of accumulated human capital.
Yet accounting sees:
Cheney. One hour. Cost.
AI makes that accumulated capability increasingly extractable.
And once capability becomes extractable, it becomes separable from the human.
That is why ownership suddenly matters so much.
From salary to licensing?
Now take individual ownership to its logical conclusion.
I own Digital Cheney.
An employer benefits from her while employing me.
Why shouldn’t it contribute towards maintaining her?
Perhaps future benefits packages contain a Digital Me allowance.
An organisation might contribute towards the cost of developing and maintaining my personally owned capability infrastructure while I work there.
In exchange, it receives clearly defined access rights.
Then I leave.
The organisation may have perfectly legitimate rights to organisational intellectual property.
But Digital Cheney herself travels with me.
Perhaps the organisation still wants access to a defined element of her capability.
Fine.
License it.
And if Digital Cheney continues producing economic value after biological Cheney has stopped providing labour, there is at least a credible argument that biological Cheney should continue participating economically in that value.
That moves us from:
organisation owns technology + purchases human time
towards:
individual owns augmented capability + organisation contracts for access and outcomes.
That intersects directly with Outcome-Based Working.
And it creates a very different interpretation of human capital.
But ownership is not one thing
There is another problem emerging as I look at this market.
Several vendors use language around ownership.
But ownership can mean very different things.
Can I control access?
Can I download my data?
Can I export the memory?
Can I export the model?
Can I move the Digital Me to another provider?
Can I prevent the provider training another model from me?
Can I commercially license my Digital Me?
Can I revoke an employer’s access?
Can I delete it entirely?
Those aren’t equivalent rights.
So one of the principles I want Bloor’s market research to test is:
Ownership ≠ Control ≠ Portability ≠ Interoperability.
A vendor saying “you own your AI” doesn’t answer the question.
It starts it.
The market is already forming
Allied is particularly interesting to this research because it has taken such an explicit position on individual versus enterprise ownership.
But it isn’t operating in an empty market.
A growing number of companies are approaching persistent human AI from different directions.
Personal AI is developing persistent personal memory and individual AI models. Its proposition centres heavily on the idea that an individual’s AI should accumulate memory and context rather than simply respond to information placed into a prompt.
Timeless AI approaches the idea through persistent AI identity: memories, voice, knowledge and personality that remain associated with the individual over time.
Sona approaches another critical element: what happens when a digital representation begins acting on behalf of a real person, and how third parties know whether they are interacting with the human or their twin.
Then there are adjacent propositions developing around expert twins, creator AI, AI coaching, enterprise employee twins, personal knowledge systems, synthetic personas and digital legacy.
Some of these systems may ultimately qualify as what I would call a Human Digital Twin.
Some may prove to be sophisticated retrieval.
Some may be extraordinarily good representations of an individual without reproducing meaningful cognition.
Some may become agents capable of acting independently.
And some may simply be enterprise automation wearing a human face.
That is exactly what I want to find out.
We therefore need two measures of Human Digital Twin maturity
Feature comparison won’t be enough.
I think Bloor needs to evaluate the emerging market across two separate axes.
The first is capability maturity.
| Level | Capability | The test |
| 1. Representation | Voice, appearance, communication style | Can it reproduce me? |
| 2. Memory | Knowledge, decisions and experience | Can it remember me? |
| 3. Cognition | Reasoning, judgement and evolution of thought | Can it understand how I think? |
| 4. Agency | Governed autonomous action | Can it responsibly act as me? |
But capability maturity is only half the assessment.
An extraordinarily capable Level 4 Digital Me with the wrong governance could be considerably more dangerous to the individual than a Level 1 avatar.
So we need a second axis.
The Human Sovereignty Test
Every proposition in this market should, I believe, be tested against a common set of questions.
Ownership: who owns the Digital Me?
Control: who determines what it can know and do?
Portability: can I take it with me?
Interoperability: can I move its accumulated capability into another technological environment?
Provenance: can I identify why it believes something about me?
Temporal integrity: does it know that something I believed in 2018 may not represent what I believe in 2026?
Contestability: can I say, No. That isn’t me?
Boundaries: can I determine what it knows, shares and acts upon?
Revocability: can previously granted access be withdrawn?
Model independence: does Digital Cheney survive if today’s technology provider disappears?
Economic participation: who receives the value generated by my digital capability?
And ultimately:
Succession: what happens to Digital Cheney when biological Cheney dies?
These aren’t peripheral governance questions.
They define the product.
This is now a market research question
And this is where I want to open the research beyond Bloor, Allied, Bersin and my own experience.
I don’t want to produce another market report based largely on product pages and vendor demonstrations.
I want to understand the thinking behind the propositions.
Because this category is being designed now.
And the assumptions being embedded now may determine the relationship between human beings, employers and AI for decades.
So before Bloor publishes its fuller market analysis of Human Digital Twins and Digital Me, I want to speak to the people building them.
Not just companies calling their product a Human Digital Twin.
If you are developing a:
Human Digital Twin.
Digital Me.
Personal AI.
Persistent personal memory.
AI self.
Cognitive twin.
Expert twin.
Employee twin.
Digital identity.
Digital legacy.
Or something you believe belongs in this emerging category…
I want to hear from you.
And I’m less interested in your demo than your thinking
Of course I want to understand what the technology does.
But I’m particularly interested in the questions underneath it.
What exactly are you trying to replicate?
Voice and personality?
Knowledge?
Memory?
Reasoning?
Judgement?
Agency?
Who owns it?
The individual?
The employer?
The platform?
Some combination?
Can it travel?
What happens when somebody changes employer?
What happens when they change technology provider?
Can it evolve?
Can your system distinguish between something I believed ten years ago and something I believe today?
Can it explain itself?
Can I ask why Digital Cheney thinks Cheney believes something?
Can I challenge it?
What happens when it gets me wrong?
What can it do without me?
Where is the boundary between augmentation and autonomous action?
What happens economically?
If my digital capability produces value while I’m not there, or after I’ve left, who gets paid?
And eventually:
What happens at retirement?
Incapacity?
Death?
Deletion?
Inheritance?
Those questions will arrive whether the market is ready for them or not.
And perhaps the biggest question is still the simplest:
Are you building something that belongs to the human, or something that captures the human for somebody else?
And I’m going to put myself through the test too
I’m not asking vendors questions I’m unwilling to test personally.
Richard Skellett already has Digital Richard.
My Digital Me is now on order with Allied Worldwide.
Which puts me in the slightly inconvenient position of becoming one of my own research subjects.
Good.
Because alongside interviewing the market, I want to document what actually happens when we attempt to build Digital Cheney.
What goes in?
What does it learn?
What does it fail to understand?
What does it infer?
What does it infer incorrectly?
Can it distinguish knowledge from opinion?
Can it recognise when my position changes?
Can it retrieve something I genuinely cannot remember?
Can it show me where that information came from?
Can I correct it?
Can I delete something?
Can I restrict access?
Can I move it?
What happens to knowledge I created before Bloor?
What happens to knowledge created with Bloor?
What happens to intellectual property belonging to another organisation?
What happens when those sources combine inside one piece of reasoning?
And perhaps the hardest question of all:
Does Digital Cheney actually feel like Cheney?
That gives the research both sides.
Vendor proposition versus lived reality.
Because the alternative deserves serious consideration
There is a dystopian version of all of this.
An organisation observes me for years.
It captures my meetings.
My writing.
My decisions.
My processes.
My judgement.
My relationships.
Eventually it creates a sufficiently useful digital representation of my professional capability.
Then I leave.
Or become expensive.
Or work fewer hours.
Or experience menopause.
Or take maternity leave.
Or become ill.
Or retire.
And the organisation says:
That’s fine. We’ve still got Cheney.
That is not augmentation.
That is extraction.
And technologically, the difference between those futures may be remarkably small.
The difference is architecture, governance, economics and ownership.
Which is precisely why these cannot be questions we answer after the technology has already been deployed.
Bersin and Skellett may therefore be debating something much bigger than Digital Twins
At first glance, the disagreement captured by the BBC looks like an intellectual-property question.
I think it’s considerably bigger.
Bersin’s model is a logical extension of the existing enterprise.
The organisation owns its systems and much of the work product created within them.
AI increases the productive capacity of employees.
Greater productivity creates greater enterprise value, some of which can flow back to employees through increased remuneration.
Skellett’s model points towards a different labour market.
The human increasingly owns an augmented capability asset and grants organisations access to it.
One reinforces the enterprise boundary.
The other potentially weakens it.
One asks:
How can the organisation preserve the capability of its people?
The other asks:
How can people preserve their capability as they move between organisations?
And my menopause-induced third question is:
How can I preserve access to my own capability as I move through different versions of myself?
All three are legitimate questions.
But they produce very different futures of work.
So who are we actually building Digital Me for?
That is now the question I think matters most.
Not:
Can we build Digital Cheney?
Increasingly, we can.
Not simply:
Will Digital Twins improve productivity?
Early implementations suggest they can.
The deeper question is:
What economic relationship do we want that capability to create between the individual and the organisation?
Because AI gives us an extraordinary opportunity.
We can preserve human knowledge.
Augment cognition.
Extend capability beyond time.
Support people through biological and life transitions.
Allow expertise to travel.
Create new forms of portfolio work.
And potentially give individuals an asset that compounds throughout an entire career.
Or we can use much the same technology to extract knowledge from workers, convert it into organisational capability and reduce our future requirement for the people who created it.
Both futures are technologically plausible.
Which one emerges will depend much less on the sophistication of the AI than on the architecture we put around it.
And that brings me back to FusionWork™.
If we’re going to reset work for the age of AI anyway, let’s do the reset properly.
Let’s decide what belongs to the human.
What belongs to the organisation.
What each should be able to access.
How value should be shared.
What rights should survive the employment relationship.
And how the accumulated intellectual capital of a human working life should be recognised.
Because we’re no longer simply deciding who owns a piece of software.
We’re deciding who owns the digital manifestation of a human working life.
And I would rather we answer that question before somebody answers it for us.
Are you building in this space?
I’m currently speaking with organisations developing Human Digital Twins, Digital Me, Personal AI, persistent personal memory, cognitive twins, AI identity and adjacent technologies as part of Bloor Research’s forthcoming market analysis.
I want to hear your proposition, but more importantly, I want to hear your thinking.
Who owns the human’s digital capability?
How portable should it be?
What does ownership actually mean technically and commercially?
How do you maintain fidelity as a person changes?
How do you distinguish memory from cognition?
Where does human IP meet organisational IP?
How should economic value be shared?
And where should the boundary sit between augmentation and replacement?
If your organisation is working on those questions, I’d like to talk to you.
Because this article is not the conclusion of the research.
It’s the invitation to participate in it.
Print Article for this blog Download here
