The biggest companies built with AI won’t be AI companies. They’ll be plumbing companies.
In February 2026, investors erased over a trillion dollars in market value from software stocks in a matter of weeks. The financial press called it the SaaSpocalypse. Analysts scrambled to explain why SaaS companies, the most reliable growth engine in technology for twenty years, were suddenly being priced for obsolescence.
The cause was simple enough to fit in a sentence: if an AI agent can do the work, you don’t need the software that was built to help a human do the work. Per-seat licensing stops making sense when the seats are empty. Klarna walked away from Salesforce and built its own AI system. Founders started replacing entire teams with coding agents. These weren’t hypotheticals. They were line items on income statements.
But the SaaSpocalypse isn’t the interesting part. Everyone is writing about the SaaSpocalypse. What almost nobody is writing about is what comes after.
Here’s what’s actually happening, stated plainly: intelligence is becoming infrastructure.
For twenty years, software was the thing you bought to run your business. CRM for sales. ERP for operations. Accounting software for money. Marketing automation for leads. Each a standalone product with a standalone price tag. An entire economy was built on the assumption that businesses would always need specialized software, and that someone would always need to sit in front of it.
Both assumptions are breaking at the same time.
The software is being commoditized because AI can now generate, operate, and maintain it. The seats are disappearing because AI doesn’t need a graphical interface — it talks directly to the systems underneath. The iShares software ETF traded nearly twenty percent below its 200-day moving average in March. Software revenue multiples have fallen from above seven to below five. The market is telling you something. Code is no longer scarce.
If code is no longer scarce, the question every builder needs to answer is: what is?
The answer is boring. Literally.
When intelligence becomes infrastructure, the bottleneck moves to the things AI cannot do. Shaking hands. Walking a job site. Tasting the food before it goes out. Sitting across from a client and reading the room. Making a judgment call when the stakes are real and the data is ambiguous. Showing up, physically, in a place where your presence matters.
These are the skills that run a plumbing company, a property management firm, a small logistics operation, a dental office. The boring businesses. The ones that never had a TechCrunch headline and never will.
They are about to have the biggest upgrade in operational capability since the internet.
Think about what it takes to run a small business today. Answer phones, respond to emails, schedule appointments, send invoices, chase payments, manage payroll, handle bookkeeping, post on social media, update your website, manage inventory, file taxes, stay compliant with regulations — and occasionally find time to do the actual work the business exists to do. Most owners spend the majority of their time on administration. The work around the work. They can’t afford to hire specialists for each function. So they do it themselves, badly and exhaustedly, or they don’t do it at all.
Now imagine all of that handled by an AI workforce. Not a chatbot. A workforce. Scheduling agents. Invoicing agents. A bookkeeping system that reconciles itself. A marketing agent that posts, responds, and adjusts without being asked. A compliance agent that tracks regulatory changes and flags what matters. An operations layer that coordinates all of it — for less than the cost of a part-time hire.
This sounds like science fiction if you haven’t been paying attention. It sounds like Tuesday if you have.
The skepticism is understandable. A plumber running a $2M operation with zero employees and an AI back office sounds like the kind of thing someone says at a conference right before they try to sell you a course.
But the same skepticism existed for every previous version of this pattern.
A single person publishing a newspaper sounded absurd until blogging existed. A single person running a global retail operation sounded absurd until Shopify existed. A single person broadcasting to millions sounded absurd until YouTube existed. Each time, a technology layer removed the operational overhead that previously required a team, and individuals with domain expertise suddenly had leverage that was reserved for organizations.
AI is doing this again, but to a category of work so broad it’s hard to overstate: administration itself. The entire back office of a business — the part that exists not because it creates value but because someone has to keep the lights on — is being automated. Not by a single product. By the convergence of language models, agent frameworks, and open-source tooling that makes it possible to assemble a custom operations layer for the cost of an API bill.
The people who benefit most from this will not be AI engineers. They will not be in Silicon Valley. They will be the contractor who’s great at building houses but drowning in paperwork. The dentist who spends more time on insurance billing than on patients. The restaurant owner who can cook but can’t market. People who have always had the domain expertise but never had the operational leverage to scale without hiring a team they couldn’t afford.
I want to be honest about the timeline. The tools are early. The orchestration is janky. The gap between what a technical person can build today and what a plumber can deploy tomorrow is still real. But that gap is closing at a speed that makes most technology transitions look glacial. Two years ago, building an AI workforce required a team of engineers. Today it requires one person willing to spend a weekend stitching together open-source tools. Next year it will require clicking a button.
We are living through a digital renaissance, and most people haven’t noticed because they’re looking in the wrong direction. They’re watching the AI companies. The interesting thing is happening behind them — in the trades, in the service businesses, in the boring industries where the hard part was never the technology. It was the overhead.
A lot of things that seemed impossibly difficult suddenly became easy. Building software. Generating content. Automating operations. Coordinating agents. Each of these was a specialized skill that took years to develop. Each is being compressed into something a motivated person can learn in a weekend. The original Renaissance happened when the printing press made knowledge accessible outside the established institutions. This one is doing the same thing to capability itself. The ability to operate a business at scale is no longer something that requires an organization. It requires a person and an internet connection.
The most interesting businesses of the next decade will be invisible to the technology press. They’ll be run by people who’ve never written a line of code, who have no opinion on transformer architecture, who could not tell you the difference between GPT and Claude. They’ll be plumbers and accountants and restaurant owners who figured out, earlier than their competitors, that the overhead that had always capped their growth was gone.
They won’t call it AI. They’ll just call it how they run their business.
And they’ll be right. The best technology disappears into the way things are done. SaaS is dead. Long live the boring business that never needed it in the first place — it just needed the work to get done.
- * *