Human capital and token capital: Satya Nadella just described augmentation
TL;DR
Satya Nadella has written down the argument we have been making for two years, in the language of strategy rather than marketing. He calls it human capital and token capital. We call it augmentation. Same idea: AI does not replace your people, it compounds them, but only if you build the loop and keep a human connecting it. The firms that turn their workflows, knowledge and judgement into agentic systems that improve with every use will pull away. Refuse, and the same AI that could have worked for you commoditises your knowledge out from under you instead.
What did Satya Nadella actually say?
On 14 June, Satya Nadella posted his thinking on the future of the firm in an AI economy. It is worth reading in full. Here is the spine of it.
This shift is different from any platform change before it. For the first time, you can build a real cognitive loop between people and machines. Not a tool that speeds up a human. A loop where the two compound.
He splits that into two kinds of capital.
"Human capital comprises the knowledge, judgment, relationships, ingenuity, and pattern recognition of its people, while token capital is the firm's AI capability it builds and owns."
Then the line that matters most:
"Human capital does not become less valuable as token capital grows. It only becomes more valuable. I believe human agency will be the driver of token capital growth... Without human direction, you have compute running in circles."
The real opportunity, he says, is not picking the best model. It is building a learning loop on top of models, where human and token capital compound together. And the line I would frame on a wall:
"You can offload a task, or even a job, but you can never offload your learning."
He calls the loop a hill climbing machine. It compounds. The companies that build it early get an advantage that is hard to copy, no matter what new model lands next month.
Why this is just augmentation in a suit
We have been saying this in plainer words since the start. The word we use is augment. It is our brand verb for a reason.
Get the AI to loop. Keep a human connecting the loops.
That is the whole game. Nadella has dressed it in the language of capital and IP, which is exactly the language a board needs to hear. But it is the same thing we tell a roofing firm drowning in quote requests or a wealth manager buried in compliance admin.
AI is augmenting. AI is not coming for your job, unless you do not use AI.
Embrace it. Take your workflows, your wisdom, your hard-won judgement, and turn them into systems that improve and loop. That is not a threat to the people who built that knowledge. It is the highest-value thing they will ever do with it.
The good news is that a lot of smart people are finally singing from the same hymn sheet. When the CEO of Microsoft and a bloke who runs a tyre shop and a property business land in the same place, the argument is settled. The only question left is whether you act on it.
The line that should change how you think
"You can offload a task, or even a job, but you can never offload your learning."
Sit with that one.
Nadella also makes a point about commoditising knowledge, and this is the part that should light a fire under you. AI can absorb expertise and make it cheap. That sounds like a threat. It is, if you sit still.
But flip it. The AI that can commoditise knowledge can commoditise it for you. It can take your knowledge, your process, your way of doing things, and make it scale at near-zero cost. That is the prize.
Without your augmentation, you do not escape the wave. You get swept under it. You are left adrift in a sea of intelligence and thinking systems that will happily do your job at a price you cannot match. The choice is not whether the sea rises. It is whether you learn to sail.
See where AI fits in your business. Free.
A 45-minute audit. We map the highest-value automations and what they're worth in time and money. No pitch, no pressure.
A real example, built last week
Let me give you proof, not theory.
I have written 189 articles about my journey into AI. Last week I turned every single one into a podcast episode. You can hear them at anaboo.ai/listen.
That was, realistically, one prompt.
Now sit with what that means.
The person who records a podcast the old way is out of a job. Bam. Gone.
But not if they augmented. Not if they learnt how to do this. Then they are not out of a job, they have a hundred of them. Imagine turning every blog on every website into a podcast summary you can play on your phone on the school run. That is real value. You just invented a new business.
Now go one further. Use the same loop to transcribe the audio and place the conversation alongside the original article. Boom. You have refreshed every article you wrote years ago and updated the lot for the agentic era. In an afternoon.
That is human capital and token capital compounding. One human with judgement, pointing the loop at the right job.
Switch the generalist, keep the veteran
Here is the part Nadella nails that most people miss.
"A company should be able to switch out a 'generalist' model without losing the 'company veteran' expertise built into their learning system. This is the key test of your control and sovereignty in the era ahead."
Spot on.
Your value does not live in the model. Models change every few weeks. Your value lives in the loop you own: your workflows, your knowledge base, your accumulated judgement, sitting in a system that improves with use and answers to you.
This is exactly what AIOS is built to do. The knowledge base holds your institutional memory and makes it queryable. The workflows get sharper every time your team uses and corrects them. And because it is model-agnostic, you can swap the engine underneath without losing the veteran on top. You keep the IP. You keep the sovereignty. The model is just the current best tool, not the thing you are betting the firm on.
That is the difference between renting intelligence and owning a compounding asset.
On regulation, late as usual
Nadella spends the back half of his post on the political economy. He warns against a world where a few models eat everything and capture all the value.
"There is no societal permission for an AI future that hollows out entire industries."
He is right, and governments are finally starting to wake up and be heard. Late, but that is usual. He draws the parallel to the first wave of globalisation, where whole industrial economies were hollowed out by outsourcing while the GDP numbers looked fine on paper. The displacement was real. We are still living with it.
The lesson for you is not to wait for the regulation. It is to make sure your firm is one that owns its loop, rather than one whose knowledge gets quietly commoditised while everyone argues about policy.
The stable equilibrium
Nadella ends on a phrase that stuck with me: the stable equilibrium we should build together.
That word matters more than it looks.
Here is where I have landed. Once you have grasped AI, and more importantly agentic AI, and workflows that run on agentic AI, and then workflows of agentic AI augmented by humans, something settles. The course is clear. The wind is behind the sails.
New models launch. Billions get announced in funding. The advance is relentless. And none of it rattles me anymore, because my business is not built on any one model. It is built on the loop. I have reached a kind of zen. A stable equilibrium.
That is the thing on offer here. Not the chase. The calm.
What to do this week
- Read Satya Nadella's post in full. Then read it again as your own to-do list, not as Microsoft news.
- Pick the single worst recurring task in your week and write down exactly how a good person does it. That document is the seed of your first loop.
- Turn that one task into an agentic workflow with a human checking the output. Watch where it gets it wrong, and feed the corrections back in. That is your token capital starting to compound.
- Decide, today, that knowledge lives in a base you own, not scattered across people's heads and inboxes. That is the difference between a firm that compounds and one that gets commoditised.
- Stop optimising for the best model. Start building the loop that survives the next ten models.
Where to from here
Credit where it is due: Satya Nadella has given the boardroom the words for what we have been building all along. Human capital and token capital, compounding through a loop you own. We just call it augmentation.
If you want that stable equilibrium for yourself, built on workflows of agentic AI augmented by humans, give us a call. We will show you the first loop worth building in your business, and the ones that can wait.
Live with passion & AI,
Brett
Running an event? Put practical AI on your stage.
Keynotes and workshops that send business owners home with a plan they can use Monday morning. No hype.
Frequently asked questions
What is the difference between human capital and token capital?
+
Human capital is the knowledge, judgement, relationships and pattern recognition your people carry. Token capital is the AI capability your firm builds and owns. The two compound each other. Human capital does not shrink as token capital grows, it gets more valuable, because someone has to point the AI at the right problems.
Will AI take my job?
+
Not on its own. AI takes the job of the person who refuses to augment. The podcast producer who learns to run the AI keeps the work and does ten times more of it. The one who waits gets passed. The dividing line is whether you build the loop or stand outside it.
What is an AI learning loop?
+
A system where your workflows, knowledge and judgement get fed into AI, the AI does the work, a human checks and corrects it, and those corrections improve the system for next time. It compounds. Every pass makes it sharper at how your business actually works.
What did Satya Nadella mean by switching the generalist without losing the company veteran?
+
He means you should be able to swap the underlying AI model, the generalist, without losing the institutional expertise your business has built on top of it, the company veteran. Your value lives in the loop you own, not in any single model. That is the real test of control.
What is a stable equilibrium in the context of AI?
+
It is the point where new model launches, funding rounds and announcements stop rattling you, because your business runs on workflows of agentic AI augmented by humans that improve no matter which model is on top. You stop chasing the news and start compounding.
How do I start building human capital and token capital in my business?
+
Pick your worst recurring task, write down exactly how a good person does it, and turn that into an agentic workflow with a human checking the output. Keep the knowledge in a base you own. That first loop is your token capital starting to compound.
Brett is a four-time founder (Darra Tyres, Gladfish, EzyTrac, Anaboo) and the operator behind AIOS, Anaboo's AI Operating System. He writes from inside the build, installing AI in his own businesses first and reporting back what actually moves the numbers. Based between Singapore, the UK and Australia.
