The Work Is Yours
AI is not going to take your job. One company is going to use AI to take your job. Unless you build first.
Everyone is talking about AI taking jobs. They're right, but they're looking at the wrong threat.
The danger is not that AI can do your job. It can. The danger is that one company will use AI to do your job and capture all the value you used to earn.
Right now, somewhere in San Francisco, a well-funded startup is building an AI that does what you do. If you're an insurance broker, someone is building the autopilot that shops carriers and fills forms without you. If you're a bookkeeper, someone is building the system that closes the books without you. If you're in HR, logistics, legal, recruiting, property management — someone is building it. They have $50 million in venture capital and a room full of engineers.
They don't know your clients. They've never done your job. They've never sat across from a business owner and explained why their coverage is wrong. They've never chased a missing W-2 at 11pm on April 14th. They've never talked a nervous hire through their first day.
But they don't need to. The AI is smart enough now. And when they finish building it, they'll sell the work you used to sell — to the same people who used to pay you.
That is the real threat. Not that AI replaces the work. That someone else uses AI to replace you, and keeps the money.
There is another path
You know the work better than any engineer in San Francisco. You know which steps break. You know which clients are difficult and why. You know what "good" looks like in your field — not from a textbook, but from fourteen years of doing it every day. That knowledge is worth something. It is worth exactly what you've been earning, and probably more.
What if you could take everything you know and turn it into an AI-operated business that you own?
Not a side project. Not a chatbot. A real business that does the work, serves customers, sends invoices, and runs under your name. Where the AI handles the repetitive parts — the document chasing, the form filling, the follow-ups, the scheduling — and you handle the judgment calls. Where you set the price. Where you own the customer relationship. Where every improvement in AI makes your business better, not someone else's.
That is what MinuteWork does.
You upload your resume. The system analyzes your experience — not your education, not your certifications, but the actual work you've done. The workflows you've run a thousand times. The pain you've watched clients endure. The follow-ups that always fall through the cracks. Then it generates a business you can build: who it serves, what it costs, who your first customers are, and the AI workforce to run it.
Your former employer might be your first customer. They already know you. They already trust your judgment. They already have the broken workflow you're about to fix. The difference is that now you're selling them the outcome instead of your hours.
Intelligence versus judgment
AI has gotten very good at intelligence work — the kind of work that has complex rules but is still fundamentally rules. Filling out insurance applications. Coding medical claims. Drafting NDAs. Reconciling bank statements. Chasing missing documents. Scheduling appointments.
Judgment is different. Judgment is knowing which client needs a call instead of an email. It's knowing when a claim feels wrong before you can explain why. It's knowing that a hire is going to work out because you've seen the pattern a hundred times.
AI is not good at judgment yet. But it's getting better. And this is the critical part: the first company to accumulate domain-specific judgment data will have an enormous advantage. Every claim processed, every policy placed, every book closed, every candidate screened — it all becomes training data for better judgment.
If a venture-backed startup captures that data, they own the future of your profession. If you capture it, you do.
The builders are already here
Anthropic recently ran a hackathon. Out of 13,000 applicants, 500 got to build for a week with Claude. The winners were a lawyer, a cardiologist, a musician, a civil engineer, and a developer.
The lawyer built a tool to streamline housing permitting in California. The cardiologist built a system that helps patients understand and act on their medical visits. The road engineer in Uganda built a tool that turns dashcam footage into infrastructure investment recommendations.
None of them needed an engineering team. None of them raised capital. They just knew the problem — deeply, personally, from years of doing the work — and the AI was smart enough to turn that knowledge into working software.
This is the part most people miss. The conversation about AI and jobs assumes there are two kinds of people: those who build AI and those who get replaced by it. But the hackathon tells a different story. The best builders aren't the ones who know the most about AI. They're the ones who know the most about the problem.
A cardiologist knows which post-visit information patients actually need. An engineer at a Silicon Valley AI lab does not. A road engineer in Uganda knows how to read dashcam footage for infrastructure decay. A machine learning researcher in London does not. A housing lawyer in California knows exactly where the permitting process breaks. A product manager at a startup does not.
The AI doesn't need you to be technical. It needs you to know what to build. That's judgment. And you already have it.
The hackathon produced tools. MinuteWork turns those tools into businesses. The cardiologist doesn't just build postvisit.ai for one hospital — they operate it as a service for every cardiology practice in Belgium. The lawyer doesn't just fix permitting for one city — they run an AI-operated permitting desk for every municipality in California. The road engineer doesn't just analyze one highway — they sell infrastructure assessment to every transportation department on the continent.
The difference between a hackathon project and a business is the operating layer: customer workspaces, billing, intake, approvals, follow-up, and the console to run it all. That's what MinuteWork provides.
The consolidation threat
In every industry, services spend dwarfs software spend. For every dollar a company pays for QuickBooks, they pay twelve dollars for an accountant. For every dollar on an ATS, they pay twenty on recruiters. For every dollar on a case management system, they pay fifty on adjusters.
When AI gets good enough to sell the work instead of the tool, whoever controls the AI captures the services budget. That is a massive transfer of value. And right now, the default path concentrates it into a handful of companies.
One company does all the insurance brokerage. One company does all the bookkeeping. One company does all the recruiting. Centralized, venture-backed, optimized for margin. The domain experts who built those industries become unnecessary.
This has happened before. Uber didn't just disrupt taxis — it turned experienced drivers into commodity labor on someone else's platform. Airbnb didn't just disrupt hotels — it made property owners dependent on a platform that takes an increasing cut. The pattern is always the same: a technology platform captures the demand side, commoditizes the supply side, and extracts margin from the middle.
AI autopilots are about to do this to professional services. Unless the professionals build first.
The operator model
MinuteWork exists so that the person who knows the work owns the business.
An insurance broker who has placed policies for fourteen years has something no startup can replicate: a network of agents who trust them, deep knowledge of which carriers actually pay claims, and an instinct for when a submission needs a phone call instead of a portal upload. That broker can launch PolicyDesk AI — an AI-operated intake and submission desk — and sell it to the small agencies they already know. Their first three customers are a phone call away.
A bookkeeper who has closed the books for fifty small businesses knows exactly which clients lose receipts, which ones need hand-holding at tax time, and which ones will pay more for a service that just handles it. That bookkeeper can launch CloseBooks AI and charge $500 a month per client to do what they used to do for $2,000 a month — except now AI handles the chasing and reconciliation, and they handle the judgment.
A logistics manager who has spent a decade keeping small carriers compliant knows every DOT deadline, every document that goes missing, every driver file that's never complete. That manager can launch CarrierWatch and serve the carriers too small for a compliance department.
None of these people need to raise venture capital. None of them need to hire engineers. None of them need to learn to code. They need to know the work — and they already do.
How it actually works
You describe what you do. Resume, LinkedIn, job description, or just plain language. MinuteWork extracts the patterns: which workflows are repetitive, which are document-heavy, which involve follow-up, which have clear buyers.
It generates ranked business ideas. Not vague startup pitches — specific, launchable operations. Who the customer is. What the price should be. Who your first ten customers are. What the AI workforce looks like: which agents handle intake, which chase documents, which build packets, which queue things for your review.
Then it builds it. The agents, the workflows, the customer workspaces, the intake forms, the landing page, the billing. You review it. You adjust it. You decide which actions need your approval and which can run autonomously. Then you start selling.
You pay nothing until your business earns money. MinuteWork takes 10% of what your customers pay you. If you earn nothing, you pay nothing.
Why this matters now
The window is short. Right now, the domain experts still have the advantage. They know the work, they have the relationships, they have the trust. The venture-backed autopilots are still building. They're raising rounds, recruiting engineers, and burning through the first eighteen months of product development.
But once they launch and start accumulating customers and data, the advantage shifts. It gets harder to compete with a company that has processed a million claims when you're just starting with your first ten.
The time to build is before the consolidation happens. Not after.
The bet
Sequoia says the next trillion-dollar company will be a software company masquerading as a services firm. They might be right. But there's another outcome — one where that value doesn't consolidate into a single company at all. Where thousands of domain experts, each serving the niche they understand best, collectively operate more of the economy than any single platform could.
Not one company closing all the books. A thousand bookkeepers, each closing the books for the clients who trust them, using AI to do it faster and cheaper than they ever could alone.
Not one company placing all the insurance. Ten thousand brokers, each running an AI-operated desk for the agencies in their network.
Not one company recruiting all the candidates. A hundred thousand recruiters, each operating an AI-powered pipeline in the industry they actually understand.
MinuteWork is the substrate that makes this possible. We build the runtime, the agents, the billing, the customer workspaces. You bring the judgment.
Someone is going to productize the work you understand. It should be you.