We are living through one of the most misunderstood paradoxes of our time.

Everyone keeps saying “AI agents will write all the code,” “software development jobs are doomed,” and “soon we’ll just tell the computer what we want and it will magically appear.”

Let’s look at what actually happened the last time humanity built machines that could produce things at unimaginable scale.


The Lesson from 1800–1950 That Everyone Forgets

Before the Industrial Revolution, making a single textile could take days of human labor. A skilled weaver working 14-hour days might produce a few yards of cloth.

Then came the spinning jenny, the water frame, the power loom, and eventually the factory system. By 1850, Britain was producing hundreds of times more cotton cloth per worker than in 1750.

Output exploded.

What happened to demand?

It didn’t stay constant. It didn’t even grow linearly.

It detonated.

Suddenly cloth was cheap enough for working-class families to own multiple shirts. Fashion emerged as an industry. New garments were invented (ready-made suits, dresses with complex patterns). Entirely new desires appeared that nobody even knew they had: curtains, upholstery, tablecloths, bedsheets in every bedroom, clothing that changed with seasons and social trends.

The cheaper and more abundant the goods became, the more uses people found for them, and the more new desires were born.

By 1900, the world consumed vastly more textiles per capita than in 1800—even though production was “automated.”

The same story repeated with steel, electricity, automobiles, chemicals, and virtually every industrialized product.

Abundance didn’t kill demand.

Abundance created demand that never existed before.


Software Is About to Do the Same Thing

Today we stand at the threshold of the same inflection point with software.

Right now, writing code is painfully slow and expensive. A decent mobile app still costs $50,000–$500,000 and six months. A custom internal tool for a mid-sized company can easily run seven figures.

AI agents—autonomous systems that can plan, write, test, deploy, and maintain code—are about to make software 10×, 100×, maybe 1,000× cheaper and faster to produce.

And exactly like cloth in 1800, the moment software becomes abundant, our appetite for it will become insatiable.

Here are just a few categories of software that barely exist today because they are too expensive, but will explode tomorrow:

  1. Hyper-personalized education
  2. Every child gets their own private tutor app that adapts in real time, speaks their dialect, knows their hobbies, and teaches physics through Minecraft-style simulations tailored to their exact interests.
  3. Personal life operating systems
  4. An agent that manages your calendar, finances, health, relationships, and learning goals in a unified way—and actually understands trade-offs instead of just syncing Google Calendar.
  5. Micro-SaaS for every niche
  6. Today we have one CRM for real estate agents. Tomorrow we’ll have 10,000 slightly different CRMs, each perfectly tailored to agents in Denver vs. Miami, luxury vs. first-time buyers, Spanish-speaking vs. Mandarin-speaking clients.
  7. Real-time adaptive enterprise software
  8. Your ERP system rewrites itself every week based on what actually works in your company this month—not what worked for the average company in 2018.
  9. Creative and entertainment tools that don’t exist yet
  10. Games that procedurally generate 100-hour storylines customized to your personality. Music production suites that finish your tracks exactly the way you would have if you had 20 years of training. Movie editors that turn your phone footage into something that looks like a $200 million blockbuster.
  11. Software embedded in the physical world
  12. Every machine, every room, every piece of clothing will have its own custom software stack because the marginal cost of writing it will approach zero.

We are not replacing a fixed pie of software needs.

We are creating an entirely new, vastly larger pie that none of us can fully imagine yet.


The New Bottlenecks Won’t Be Code

They will be:

  • Vision and ambition (what should exist that doesn’t yet?)
  • Taste and judgment (which of the million generated options is actually good?)
  • Integration and orchestration (making 50 AI agents work together without creating chaos)
  • Trust and verification (how do you know the agent didn’t subtly hallucinate a security vulnerability?)
  • Human experience design (the last 1% that makes software feel magical instead of just correct)

These are the new scarce resources—and the people who can provide them will be more valuable than ever.


The Bottom Line

The Industrial Revolution didn’t end the need for human creativity, design, entrepreneurship, or labor.

It multiplied it by orders of magnitude.

The AI Agent Revolution won’t end software development.

It will be the spinning jenny of the mind.

And just like Manchester in 1840, the next decade is going to be chaotic, noisy, wildly unequal, morally confusing—and absolutely thrilling for anyone who is ready to build.

The cloth is about to become infinitely cheap.

Now go invent fashion.