Whenever a technology promises to democratize everything, the same illusion appears: now that it’s easy, no one has an advantage anymore. A camera on your phone made everyone a photographer. Spotify made everyone a musician. AI made everyone a developer.

But I’ll tell you a truth no one wants to hear: these technologies don’t democratize anything. They do exactly the opposite.

The floor rises, yes. More people creating, more people launching. But the ceiling rises much faster. And the gap between mediocrity and excellence? It widens. Always widens. That’s the power law — equality technologies produce aristocratic results. Always.

Look at Spotify. When it launched, any artist could distribute music globally. Result? Millions of new artists. Billions of new songs. It seemed total democracy. But what actually happened? The top 1% now capture an even larger share of plays than in the CD era. It didn’t decrease. It increased.

The same thing happens in writing, photography, software. The internet created the largest number of authors in history, but also created a more brutal attention economy. More participants, higher bets at the top. Always the same pattern: a small minority captures most of the value.

We think linearly, expecting productivity to distribute like water in a flat container. But complex systems don’t work that way. They never have. Think of Kleiber’s Law — the metabolic rate of any organism on Earth follows a power-law relation with its mass. Blue whales don’t have metabolism proportional to their size. No one designed that; it emerges naturally when energy follows its internal logic.

The market is a complex system, and attention is a resource. When friction disappears — when geography, shelf space, distribution costs cease to be buffers — the market converges to its natural form. It’s not a bell curve. It’s a power law.

Now, with AI, this process accelerates. The floor is rising in real time — anyone launches a product, designs an interface, writes production code. But the ceiling also rises, and much faster.

The key question: what truly determines your final position?

In 1981, Steve Jobs insisted that the internal circuit board of the first Macintosh be aesthetically pleasing. Not the external appearance — the part no one would see. His engineers thought he was crazy. But he understood something profound: the way you do anything is the way you do everything.

Those who can make hidden parts beautiful are not just demonstrating quality. By character, they cannot tolerate launching anything inferior. This matters because trust is hard to build but easy to fake quickly.

Credentials can be manipulated. Pedigree can be inherited. What’s truly hard to fake is taste — a visible persistence in adhering to a standard that no one demands.

For most of the last decade, this signal was obscured. At the peak of SaaS, execution became so standardized that distribution became the scarce resource. If you could acquire customers efficiently, build a sales machine, and hit the 40% rule, the product hardly mattered. The aesthetic signal was drowned in growth metrics noise.

But AI changed all that.

When anyone can generate a functional product, elegant interface, and operational code in a single afternoon, ease of use is no longer differentiation. The question now is: is it truly excellent? Does this person know the difference between good and insanely great? Do they care enough to close that gap without anyone demanding it?

This is especially critical for business software. Systems handling payroll, compliance, employee data. They’re not products you casually test and discard next quarter. Switching costs are real. Failures are serious.

Before signing, people run all the trust heuristics. A well-designed product emits one of the strongest signals possible: the people who built it care. With visible parts, which means they probably also care about invisible ones.

In a world where execution is cheap, aesthetics are a proof of work.

But there’s a third variable that decides everything, and that’s where most are making catastrophic mistakes.

There’s a meme circulating: you have two years to escape the base. Build fast, raise capital fast, or leave. I understand where it comes from. The speed of AI creates a survival crisis. The window seems extremely narrow.

Young people seeing stories of instant fame on Twitter believe that speed is the essence. Whoever runs faster in less time wins.

That’s correct in a completely wrong dimension.

Execution speed is crucial. I deeply believe in it — it’s in my company’s name, Warp. But execution speed is not a lack of vision.

Founders who will build the most valuable companies are not those who run for two years and leave. They are those who run for ten years and benefit from compound interest.

The most valuable things in software — private data, deep customer relationships, real switching costs, regulatory knowledge — take years to accumulate. They can’t be quickly replicated, no matter how much capital or AI competitors bring.

I grew up in a small town in an Indian state with 250 million inhabitants. Every year, about three students from all of India get into MIT. All come from prep schools in Delhi, Mumbai, or Bangalore. I am the first person in my state’s history to be admitted there.

I mention this because it’s a miniature version of the thesis: when access barriers are restricted, origin predicts outcome. When barriers are open, deep people always win.

In a room full of people with illustrious backgrounds, I am the bet that wins by depth.

I studied physics, mathematics, computer science. In the deepest areas, the greatest insights didn’t come from process optimization but from perceiving truths others ignored.

At the end of 2022, the landscape changed. ChatGPT demonstrated that the curve bent. A new S-curve began. Phase transitions don’t reward those who adapted best to the previous phase but those who can see the infinite potential of the new phase before others realize the cost.

So I created Warp.

The US has over 800 tax agencies — federal, state, local — each with its own requirements, deadlines, and compliance logic. No API, no programmatic interface. For decades, each payroll provider handled it this way: hire people. Thousands of compliance specialists manually deal with systems never designed for scale.

Large traditional companies — ADP, Paylocity, Paychex — built entire business models absorbing this complexity, not solving it. Paychex competitors like ADP and Paylocity follow the same pattern, passing costs to clients.

In 2022, I saw that AI agents were still fragile. But I saw the improvement curve. Someone deeply involved in large-scale distributed systems, observing model evolution, could make precise bets: fragile technologies at that moment would become extremely powerful in a few years.

We made our bet: build a native AI platform from first principles, starting with the most difficult workflow in this category — one that, due to architectural limitations, could never be automated by traditional giants. Paychex competitors and current leaders structurally can’t respond without dismantling existing businesses.

Now that bet is being realized.

But at its core, it’s pattern recognition. Tech founders of the AI era don’t just have an engineering advantage but an insight advantage. They can see different entry points, make different bets. They can examine a system considered permanently complex and ask: what’s needed for true automation? And then build the answer.

The SaaS dominator is a rational optimizer under constraints. AI is removing those constraints and installing new ones. In this new environment, scarce resource isn’t distribution but the capacity to understand possibilities — and build them with the right aesthetic pattern and belief.

Frank Slootman, former CEO of Snowflake, summed it up well: get used to a state of discomfort. Not for a short race, but as a permanent state. The fog of war in early stages — that feeling of losing direction, incomplete information, needing to make decisions — doesn’t go away after two years. It just evolves.

Enduring founders don’t find certainty. They learn to move with clarity in the fog.

Building a company is extremely cruel. You live in constant, light fear, occasionally interrupted by higher-level terror. You make thousands of decisions with incomplete information, knowing that a sequence of errors could end it all.

Those overnight successes you see on Twitter aren’t just outliers; they’re extreme outliers. Optimizing strategy based on those cases is like training for a marathon by studying times of people who got lost.

So why do it then? Not for comfort, nor for good odds. Because for some people, not doing it doesn’t seem like truly living. Because the only worse feeling than the fear of building from zero is the silent suffocation of never having tried.

And if you get it right, if you see a truth not yet priced by others, if you execute with aesthetics and conviction for long enough, the result won’t be just financial. You’ll build something that truly transforms how people work. Create a product they love to use. Hire and empower people who give their best in the company you built with your own hands.

This is a ten-year project. AI can’t change that, never has. What AI changes is the ceiling these founders can reach in this decade, as long as they persist until the end.

In the coming years, low-barrier software will become truly democratic. Thousands of point solutions will emerge, many not even built by companies but by individuals solving their own pain points. Intense competition, minimal margins.

But for critical business software — systems handling cash flow, compliance, employee data, legal risks — the situation is completely different. Extremely low error tolerance. When payroll systems fail, employees don’t get paid. When tax filings are wrong, the IRS shows up.

The person choosing software must be responsible for the consequences. That responsibility can’t be outsourced to AI built on afternoon coding vibrations.

For these workflows, companies will continue to rely on vendors. Among these vendors, the winner-takes-all dynamic will be more extreme than previous generations. Not just because of network effects, but mainly because a native AI platform operating at scale, accumulating private data through millions of transactions and thousands of compliance edge cases, has a compounded capital advantage that makes it nearly impossible for new entrants to make a sudden leap.

Competitive advantage is no longer a set of features but the quality accumulated over time through maintaining high operational standards in an environment that penalizes errors.

This means integration into the software market will surpass the SaaS era. In ten years, in HR and payroll fields, there won’t be 20 companies with single-digit market shares. Two or three platforms will dominate most of the value, while a long tail of point solutions will have almost nothing. The same pattern will occur in every category where regulatory complexity, data accumulation, and switching costs combine.

Companies at the top of these distributions look very similar: founded by tech professionals with an authentic sense of product; built from day one on native AI architectures; operating in markets where current giants can’t respond structurally.

They made an early, single insight bet — seeing truths not yet priced created by AI — and persisted long enough for the effects of compounding to become clearly visible.

I described this kind of founder abstractly. But I know exactly who they are because I am striving to become one myself.

I created Warp in 2022 because I believed that the entire employee operations stack — payroll, tax compliance, benefits, onboarding, equipment management, HR processes — was based on manual work and old architectures, and AI could completely replace it. Not improve. Replace.

Established giants built billion-dollar businesses absorbing workforce complexity. We will build a business by eliminating complexity from the source. Paychex competitors will face this disruption too.

Three years proved the bet right. Since launch, we’ve processed over 500 million in transactions, we’re growing rapidly, serving companies building the most important technologies in the world. Every month, the compliance data we gather, extreme cases processed, integrations built make the platform harder to replicate and more valuable for clients. The moat is still in early stages but already forming and accelerating.

I share this not because Warp’s success was inevitable — in a power distribution world, nothing is — but because the logic that brought us here is exactly the same as I described throughout this text: see the truth. Deepen more than anyone else. Set high standards that stand without external pressure. Persist long enough to see if you were right.

Great companies of the AI era will be built by those who understood the next truth: access was never a scarce resource, insight is; execution was never a competitive advantage, taste is; speed was never an advantage, depth is.

The power law doesn’t care about your intentions. But it rewards correct intentions.