Of all the places I’ve ever done gongfu tea, my very favorite is my uncle’s living room. A true tea cI did not set out to become a health entrepreneur. Health first entered my working life through a spreadsheet.
In the early 2010s, I was running a company and watching one line in our costs rise faster than almost anything else. At one point, our employee healthcare costs were increasing by roughly 25 per cent a year. I was used to costs moving. Salaries rose, technology changed, suppliers negotiated and markets did what markets do. But this was different. The number kept getting larger without anybody being able to explain, in a way I found particularly satisfying, what we were getting in return.
So I began asking why.
The more I looked, the stranger the American healthcare system appeared. At its best, acute medicine is extraordinary. If I am badly injured, have a heart attack or need complex surgery, I would rather be in the hands of modern medicine than at almost any other point in human history. We have become remarkably good at intervening when something has gone seriously wrong.
But much of the cost was not coming from isolated emergencies. It was associated with people living for years with chronic and mental health conditions. Today, US healthcare spending has reached $5.3 trillion a year, and the Centers for Disease Control and Prevention says that 90 per cent of that expenditure is for people with chronic and mental health conditions. That does not mean every dollar is preventable, or that chronic disease is simply the result of poor choices. It does mean the system is carrying an enormous long-term burden.
What struck me then was how far the incentives leaned downstream. There was a sophisticated system waiting after somebody became ill, but far less coherence around helping that person remain well in the first place. We had built an astonishing repair industry. I started to wonder what a serious maintenance system for human beings might look like.
It was not difficult to see why the system had evolved that way. Acute episodes are visible. Tests, procedures and appointments can be recorded and reimbursed. Prevention is much less tidy. The action may happen at home rather than in a clinic, the benefit may not appear for years, and the employer or insurer paying today may not be the one that eventually saves the money. Traditional fee-for-service models also rewarded activity more readily than long-term health. Nobody needed to design a bad system deliberately. A collection of understandable incentives could do the job perfectly well.
That question was still rather abstract until James had his accident.
When health stopped being a line item
James’s accident shocked me. His life was disrupted, and the process of healing was not a neat clinical episode with a beginning and an end. It became a journey that affected his independence, his confidence, his relationships and the ordinary rhythms of his days.
The accident was not a lifestyle choice, of course. That is precisely why it changed my thinking. Health is never a tidy morality play in which disciplined people remain well and everybody else receives a lesson. Luck matters. Genetics matter. Accidents happen. Illness can arrive despite somebody doing everything apparently right.
But watching James rebuild also showed me that a person is not merely the passive recipient of care. What happened in the clinic mattered enormously, but so did what happened between appointments: what he understood, what he practised, how he responded to setbacks and whether he could see a reason to keep going when progress was slow.
I began to think more seriously about personal responsibility. I still believe in it, although I use the phrase carefully now. Responsibility without context quickly becomes blame, and blame is a remarkably poor health strategy. People do not make decisions in a vacuum. Money, time, stress, education, pain, neighbourhoods, work, family and the options immediately around them all shape what is realistically possible.
The more useful meaning of responsibility is agency. What can I influence from where I am? What can the people around me make easier? What information would help? What support would turn a good intention into something I could actually sustain?
Those questions moved health from a company expense to a problem I wanted to understand.
Knowing what to do is not the same as doing it
At first, I assumed education was the missing piece. Give people better information about food, movement, sleep and stress, and surely they would make better decisions.
Education matters. But the more time I spent in health, the more obvious it became that information is necessary and wildly insufficient. Most people already know, at least broadly, that moving regularly is better than sitting all day, that sleep matters and that an apple has certain advantages over a packet of biscuits. The problem is not that the entire population has somehow missed the existence of vegetables.
The problem is that health information competes with everything else in a person’s life.
It competes with an early meeting, a late shift, a child who needs collecting, an ageing parent, a difficult marriage, two jobs, a short night of sleep and the simple desire to stop making decisions for five minutes. Human beings lean towards comfort because life is often uncomfortable enough already. I do not think that makes us weak. It makes us human.
It also competes with extremely capable commercial systems. I came from marketing and technology, so I knew how much intelligence goes into shaping behaviour. Products are placed, priced, packaged, promoted and continuously refined to reduce friction. The unhealthy choice is often not merely available. It has been made convenient, familiar and emotionally rewarding by people who are very good at their jobs.
Then health professionals arrive and tell an exhausted person to exercise more discipline.
That seems an uneven contest.
The reason has to be stronger than the friction
Information can tell us what to do. It rarely tells us why we care enough to keep doing it.
Everybody has a why, although it is not always strong enough or close enough to overcome today’s friction. We may say we want to live longer, feel better or avoid disease, but those are distant and slightly bloodless ambitions when compared with the immediate pleasure of staying on the sofa or ordering something easy for dinner.
Then the why changes. A parent dies earlier than expected. A child becomes ill. Pain arrives in the body. A doctor uses a word that makes the room suddenly feel smaller. The future stops being theoretical, and behaviour that seemed impossible a month earlier can begin to feel urgent.
I have never been comfortable with the idea that we should wait for fear, grief or pain to supply motivation. The more interesting question is whether people can find a positive reason before crisis finds one for them. Perhaps it is playing with grandchildren, travelling with a partner, continuing to work, remaining independent or, in my slightly eccentric case, skiing and playing tennis on my hundredth birthday.
A useful why is not a slogan printed above a gym mirror. It is a picture of a future somebody genuinely wants to inhabit.
But even a powerful reason is not enough on its own. If the path towards it is relentlessly inconvenient, confusing or lonely, motivation is slowly spent. We need to pair meaning with design.
The decisions too small to notice
Health is shaped through an extraordinary number of ordinary decisions. Do I walk or take the car? Do I remain in the chair for eight hours or stand and move regularly? Do I cook something or reach for something processed? Do I go to bed or watch one more episode? Do I call a friend or continue scrolling through the lives of people I do not know?
Each decision is almost trivial. Their accumulated direction is not.
That is one of the difficulties with prevention. The consequences are delayed. A walk today does not transform a life by tomorrow morning, just as one poor meal does not create a chronic condition by Friday. The signal is quiet, which makes it easy to ignore. Months and years later, the effect of repeated choices can become much clearer.
Where prevention is possible, it is generally easier to act before a condition has become established than to reverse years of accumulated biology and behaviour. Yet the system often asks people to make the harder choice now in exchange for a benefit they may not feel for decades.
Technology looked as though it might solve that feedback problem. A wearable could make movement visible. A photograph could create a record of a meal without an elaborate food diary. A dashboard could show patterns that memory missed. For a technology entrepreneur, it was an irresistible proposition: make the invisible visible, shorten the feedback loop and help people learn from their own lives.
So in 2014, we built OffTheScale.
What OffTheScale taught me
We were early. Wearables were still novel, taking photographs of meals as a health tool felt more experimental than it does now, and the phrase “digital health” had not yet been attached to quite so many things.
OffTheScale combined technology with in-person support. We used wearable data to understand movement and meal images to make eating patterns easier to discuss. But we also brought people together for education, community and a shared sense of purpose. Thousands of people went through the programme, and we saw encouraging results.
The data mattered, but the more important lesson was what the data could not do by itself.
A device could show that somebody had stopped moving. It could not necessarily tell us that her mother was ill, she was barely sleeping and a daily walk had fallen to the bottom of a life already carrying too much. A meal photograph could show what somebody ate. It could not explain the family ritual, financial pressure or moment of comfort wrapped around it. A chart could record a setback. It could not decide whether the person looking at it experienced that setback as useful information or as evidence that he had failed again.
That is where the human being changed the experience. A good coach could interpret rather than merely report. A group could make a new behaviour feel socially normal. Education could reduce confusion. Conversation could reconnect a difficult week to the reason somebody started.
I had assumed the in-person element supported the technology. I gradually realised I had it backwards. The technology supported the person, and the people supported the change.
The hybrid was not a compromise. It was the product.
A framework for change
That experience left me with a simple way of thinking about behaviour change. Sustainable change usually needs four things working together.
First, a reason that matters. Not a generic instruction to be healthier, but a future the person can see and wants to protect.
Second, an easier next step. The environment has to reduce friction around the better decision rather than relying on an endless supply of willpower.
Third, useful feedback. Data should make progress and patterns easier to understand, not give somebody another dashboard to feel guilty about ignoring.
And fourth, a relationship. That might be a coach, clinician, colleague, friend, peer group or community. Somebody, or some group, helps translate complexity, offers perspective and makes continuing feel possible.
The four elements reinforce one another. A reason gives direction. Design makes action more likely. Feedback helps us learn. Relationship helps us recover and continue.
Remove any one of them and the system becomes less stable. Motivation without an easier path becomes exhausting. Convenience without meaning rarely survives novelty. Data without interpretation becomes noise. Support without agency becomes dependence.
This is the framework that began moving me from technology entrepreneur to health entrepreneur. I stopped asking only, “What can the technology do?” and started asking, “What would help this person change?”
It also changes where I think we should begin. Before buying another device or enrolling in another programme, ask what future matters enough to make the effort worthwhile. Find one repeated decision that points towards it. Look for the friction making that decision harder, decide what feedback would genuinely help and identify who could make the process feel less solitary. The answer may still involve an app, a wearable, a clinician or a coach. It may also be a walk with a friend, food prepared before a difficult week or a phone left outside the bedroom. The tool should follow the human problem, not define it.
What AI changes, and what it may not
The technology available now makes what we built in 2014 look rather primitive. Wearables collect richer data. Phones can interpret images. Artificial intelligence can identify patterns, personalise education, summarise complex information and provide support at any hour without becoming tired or impatient.
It would be nostalgic, and scientifically unwise, to insist that only a human coach can help someone change. In a 2025 randomised clinical trial, an AI-led diabetes-prevention programme was non-inferior to a human-led programme on a defined set of outcomes after twelve months. That is important evidence, and I suspect it is the beginning rather than the end of what machines will be able to do.
But I am not yet convinced that the entire challenge can be reduced to delivering the correct prompt at the correct moment.
Health change takes place inside a life. It is entangled with identity, relationships, work, money, grief, confidence and belonging. The hard moment is often not knowing whether to walk after dinner. It is continuing after the week in which everything went wrong, or changing a behaviour that quietly holds a family or friendship together.
Perhaps AI will become extremely good at understanding those contexts too. Perhaps in ten years a physical robot will inspect my refrigerator, challenge my excuses and accompany me on a walk. I may find that both useful and faintly irritating.
I may be wrong about where the boundary will settle. But for now, I believe we still need a human in the loop, even if that human is not always a professional coach. We need connection, interpretation and the feeling that somebody understands the life around the data.
More importantly, we need to keep the human at the centre of the loop. The purpose of health technology is not to produce a perfectly compliant user. It is to help a person preserve more energy, agency and capability for the life that matters to them.
The questions at the heart of my journey
I once thought rising healthcare costs were primarily an economic problem. Then James’s accident made them human. OffTheScale made them behavioural. Technology made them measurable. The years since have made them systemic.
I still believe in personal accountability, but not as an excuse for badly designed environments or fragmented care. I still believe in education, but not in the fantasy that information automatically becomes action. I believe technology can make health more understandable and support better decisions, but I no longer confuse measurement with change.
The opportunity is to build systems that respect all of those truths at once.
That means helping people discover a reason before crisis supplies one. It means making the healthier decision less burdensome in homes, workplaces and communities. It means turning data into a small number of useful choices. And it means recognising that change is easier when somebody does not feel they are attempting it alone.
These questions have been at the heart of my journey from technology entrepreneur to health entrepreneur. They also sit behind much of how I now think about healthier homes, healthier communities and Human Energy Intelligence: not as separate products or projects, but as different ways of helping people connect the conditions of their lives with the energy and capability they want for the future.
I do not think we need to choose between technology and humanity. I think the interesting work is designing each to make the other more useful.
The Bigger Question
If technology can increasingly tell us what to do, how do we build health systems that also help us care enough, and feel supported enough, to keep doing it?
A little of the evidence behind this essay
This is a personal account of how my thinking developed, not a claim that individual behaviour explains every chronic condition or that every illness can be prevented. Genetics, social and economic conditions, environmental exposures, access to care and chance all matter.
The Centers for Medicare & Medicaid Services estimates that US healthcare spending reached $5.3 trillion in 2024, or 18 per cent of the economy. The CDC reports that 90 per cent of annual health expenditure is for people with chronic and mental health conditions. That wording matters: expenditure on people with these conditions is not the same as a calculation that 90 per cent of all spending was caused by preventable behaviour.
Evidence on technology and human support is evolving. A 2019 randomised trial found that electronic health-record tools plus personalised coaching helped people maintain more weight loss over two years than tracking tools alone. A 2024 trial found that adding adaptive telephone coaching improved outcomes for people responding poorly to an online weight-loss programme. In 2025, however, a randomised trial found an AI-led diabetes-prevention programme non-inferior to a human-led programme on its composite twelve-month outcome. Together, those findings suggest that technology can do more than sceptics sometimes allow, while the best role and timing of human support remain important design questions.
Sources:
Centers for Medicare & Medicaid Services, National Health Expenditure Accounts, 2024.
