- Flywheel Insights
- Posts
- How AI will reshape work
There is a lot of talk about how AI will change industries and society.
AI will impact the workplace. But not in the way people think.
On one extreme, people believe that labor will become redundant.
The other extreme thinks that AI is overhyped.
I think the reality will be more nuanced: AI will reshape work in a barbell pattern, creating a few very large winners and many smaller businesses.
History has shown us how technological advancement has created asymmetric outcomes.
Some of the technological advancements have been like a tide that raises all boats (to varying degrees).
In other cases, technology exacerbates inequality and produces a few very large winners. These winners are those who reap the largest share of the gains.
The benefits of electricity and household appliances were shared more equally than those of advanced military technology and the internet. Consumers were the biggest beneficiaries of the former, while governments and corporations were the biggest beneficiaries of the latter.
I believe AI will impact society in a barbell-shaped way.
There will be a few multi-trillion dollar companies on one end and millions of smaller software or tech-enabled service companies on the other. My main conclusion is that micro entrepreneurship will explode, creating countless smaller businesses.
Micro-entrepreneurship will be the norm:
Countless players will emerge because individuals can now create their own digital products and reach markets much faster and cheaper with AI.
This is inevitable because starting a business has never been easier or cheaper.
Let me share my personal experience with this.
I have no experience or education in coding.
Yet, last month, I built an interactive website using AI for roughly $60 in tokens.
I completed the website in a day.
More recently, I spent less than two weeks building a portfolio management tool for the company I work for.
A few years ago, I would have had to pay a web designer at least $5,000 and wait 2-4 weeks to build a website. The portfolio management tool would have cost me more than $30,000 and taken months to complete.
A key benefit of AI is that it shortens the learning curve. You can learn new and efficient ways to do things. You can get immediate feedback or solutions about processes.
More people will use AI to improve their productivity.
And not just in their day job, but also for themselves as freelancers, solopreneurs, or founders.
A Historical Analogy:
This pattern of independent entrepreneurship is not new.
Before the industrial revolution (particularly across the Middle East and parts of Asia), most economic activity was conducted by independent merchants, often with just a few people working for them.
A 'business' was the personality and reputation of the entrepreneur. The corporation, as the dominant organizational form of commerce, is a relatively newer concept.
There were two forces behind the Industrial Revolution: economies of scale and the division of labor. Higher fixed costs could produce more at a lower cost per unit.
To be competitive, corporations had to be sufficiently large and employ many people specializing in narrow tasks. Because task switching creates inefficiencies, labor would ideally do one thing many times over. By doing this, companies, often industrial, were able to become large and achieve minimum efficient scale.
The highest levels of scale in the age of AI will belong to a limited set of players, while most new value will be created by smaller operators.
The micro-entrepreneur will sit on top of the technology stack created by IT infrastructure providers. He or she will rent the intelligence and deliver work that would have required a whole team a few years ago.
Technology has reduced the cost of communications, and labor markets have become global. So hiring people (full-time or as contractors) in low-cost markets is becoming the norm increasingly. Micro-entrepreneurs will run businesses with a handful of employees and contractors. These workers are now cheaper to hire and more flexible to work with.
Cheap and accessible technology + cheap labor = explosion of micro entrepreneurs.
Medvi: $1b in sales and 1 full time employee using AI?
Matthew Gallagher launched Medvi, a telehealth startup, with just $20,000 and AI tools.
In 2025, its first full year, the company generated $401 million in sales and served more than 250,000 customers.
It is projected to do $1.8 billion in revenue this year. Gallagher hired his brother and some contractors.
Most micro-entrepreneurs will start out by offering services with digital distribution.
That will naturally evolve into productized services and eventually digital products.
These services will revolve around the entrepreneurs' key skills or interests, or both.
This will create many unique service offerings directed towards niche customers or use cases.
For example, a business development professional at a family office can become an independent placement agent.
An associate architect can go solo.
The same logic applies to engineers, analysts, designers, and many other roles.
They will use content and personal relationships to build a reputation around their personal brand.
They will also be cheaper to hire than legacy businesses, and still make more money.
Since people trust other people more than they do corporations, relationships with these micro-entrepreneurs can be stickier than with traditional vendors.
Having a personal brand around a niche (or local) product or service has the potential to be very successful.
Technology makes it easier to become visible, accessible, and handle administrative work.
The result?
Instead of being individual contributors in a company, knowledge workers will operate independently, which is the core shift I expect AI to accelerate.
Corporations will hire these independent workers and hire a larger mix of contractors. This is better than relying on full-time employees who come with fixed contracts and liabilities.
Labor within companies will become more variable (rather than 100% fixed).
The status quo is not enough:
I don't see why the workplace status quo would stay the same - there is less of a reason for it.
The promise of the post-industrial era no longer exists.
Let me explain.
The industrial revolution created the concept of the middle class. People could live a comfortable life working as employees and by selling their labor to a corporation.
But the days of being able to buy a house, have two cars, and raise a family on a single salary are vanishing.
Real wages have been compressing.
People are expected to go into debt to buy a house, get a car, or afford an education.
Real wages have been stagnat in the United States during 1970s.
Moreover, employee productivity has grown much faster than employee compensation.
Technological advancements means that people have the ability to start their own micro-businesses and keep more of the productivity gains for themselves.
As industrial productivity grows, workers have been moving from factories to offices. But these office-laden white-collar workers now spend a large part of their careers doing boring administrative work. The type of work that can be automated with AI.
Additionally, corporations have been slowing down hiring for entry-level positions. These companies have spoken about maximizing efficiency by using AI.
This results in youth unemployment being roughly 2-2.5x the overall rate.
The stakes of going independent are lower than they used to be and the risks of leaving the corporate system is lower than its ever been.
As a result, more people will sell their services as independent contractors.
Marriage and home ownership, a proxy for wealth, has been in decline since 1960 for the <30 year old age group.
Vocational work is back in fashion:
Knowledge work is not the only kind of work being reshaped.
The historical advantages of a college degree are eroding.
Recent graduates in the US now face unemployment rates above the overall labor market.
This is a reversal of a century-long pattern.
Meanwhile, vocational and trade work has been growing, in terms of employment and compensation.
In the US, enrollment in trade schools is rising while four-year college enrollment is declining.
Median wages for electricians, plumbers, HVAC technicians and welders have grown faster than entry-level wages in many white-collar fields.
While this trend is mainly seen in developed economies, it can signal how things are likely to play out.
There is a simple reason for the growing importance of the trades.
AI is good at knowledge work and poor at physical work. The job of a plumber dealing with a variety of pipes in tight spaces is not in any immediate danger. A junior analyst writing a first draft of a memo may have less job security.
The barbell shows up in the market for trades-related businesses too.
Large facilities-management and trade roll-ups are consolidating at the top. Independent tradespeople running AI-augmented one-person shops are emerging at the bottom.
Domain expertise becomes a bigger advantage:
If general capability is becoming cheap, the question is what stays valuable.
The answer is deep, vertical-specific domain knowledge.
AI is excellent at general analysis.
It is much weaker in specific operational knowledge of a regulated industry, complex and bespoke workflows, and the demands of a particular type of customer.
Someone who understands hotel operations in the Middle East, or healthcare claims in a specific jurisdiction, becomes more valuable.
Domain expertise is becoming more valuable relative to technical skill.
Someone with deep domain expertise and zero programming background can now build functional tools through natural language.
Code is not just cheaper to produce. The number of people who can produce software has expanded by orders of magnitude.
The person who deeply understands their area of expertise and can build a tool with AI is more dangerous than the engineer who can build anything but does not know what to build.
This is also why the software giants at the top of the barbell can stay durable.
Their advantage is not the software itself. It is decades of accumulated workflow knowledge, proprietary data, and embedded customer relationships.
Individuals get the same kind of moat, on a smaller scale, by going deep into a specific vertical and pairing that knowledge with AI.
What this means for you:
If you are a young professional (0–5 years of experience): Pick a vertical and go deep. Generic competence is a commodity. Start a small AI side project this month, not next year. The cost of experimenting has collapsed, and what you learn from shipping something small.
If you are a mid-career professional (5–10 years of experience): Your decade of domain context is now more valuable than it was 2-3 years ago. Audit which parts of your job is repetitive and can be augmented with AI. Your work is becoming editorial, and the skills that matter revolve around process design, judgment and accountability.
If you run a business: Experiment with offshore and AI-augmented talent now. Embed AI into every workflow that is manual and repetitive. Get AI to support with analysis and reporting to act as a devils advocate or generate ideas for process improvement. And, most importantly, do not stop hiring juniors. The temptation to cut entry-level roles entirely seems real, but you are also cutting your company’s future leaders and doing a disservice to society. Hire more carefully and and train them well.
Concluding remarks:
The corporation isn't disappearing, but its monopoly on professional life is.
For three generations, knowledge workers were promised stable employment and comfortable means in exchange for stable output.
That whole paradigm is coming into question because of AI.
The structure of the corporation and work will shift. There will be more independent 'micro-entrepreneurs' who sell their services to people and companies. They successful ones will earn more and have more flexibility in how they manage their life.
How will you prepare for the current paradigm shift?