The Next Trillion-Dollar Company Won't Sell Software. It Will Sell the Work.

In Brief

The next trillion-dollar company won't sell software tools. It will sell completed work.

Julien Bek, a partner at Sequoia Capital, published an essay mapping the shift from AI copilots to AI autopilots, identifying the technology's real target: the $4 to $5 trillion white-collar, knowledge-based workforce.

The implications ripple across every professional services industry, from accounting to consulting to insurance. And the transition is already underway.

What Happened

For years, AI companies have been building copilots or tools that work alongside humans to make them more efficient. A Sequoia Capital partner now argues that era is ending. The next wave of AI companies won't sell tools at all. They'll sell the finished work product.

I explored this idea with SmarterX and Marketing AI Institute founder and CEO Paul Roetzer on Episode 201 of The Artificial Intelligence Show.

The core argument. Sequoia Capital partner Julien Bek published an essay titled, "Services: The New Software" that reframes how to think about the AI industry's trajectory. His main argument is that the previous generation of AI companies built copilots, tools that work alongside humans. But models have gotten smart enough that the next generation will build autopilots or companies that sell the completed work directly to the buyer.

Intelligence work vs. Judgment work. Bek draws a crucial distinction between two types of knowledge work: Intelligence work, which is complex but ultimately rule-based. Its tasks follow patterns, even if those patterns are sophisticated. And judgment work, which requires experience, instinct, and taste that is built over years. His claim is that AI has crossed the threshold to handle pure intelligence work autonomously, and it's rapidly closing the gap on judgment.

"Right now you need copilots that work with the humans because the human still has the experience, the expertise, the judgment, the taste that knows what to do next.

"What he's saying is once the models get enough training in specific domains, then they get judgment, too. And maybe we don't need the human to have the judgment anymore and it can become an autopilot."

—Paul Roetzer, founder and CEO of SmarterX

The Key Numbers

• $1 spent on software = $6 spent on services. 

• 49.7% - Percentage of software engineering in which agents are already deployed.
• 9% - Percentage of back-office automation in which agents are deployed
• 134,000 - The number of finance and insurance job openings in December, the lowest number since February 2012.  
• 1.9% - Job opening rate for finance and insurance workers, the lowest rate recorded this century

Why Software was Never the Target

Silicon Valley has always known the labor market is the biggest prize for AI. 

Revenue in the software industry in the U.S. is about $300 to $500 billion annually. Annual wages for knowledge workers or people who use computers for a living,is $4 to $5 trillion. So the knowledge labor market is literally 10x the software revenue industry.

"It's always been an inevitability that's what Silicon Valley would build towards, what VCs would fund: Companies that went after the much larger total addressable market."

—Paul Roetzer, founder and CEO of SmarterX

The existential question every AI founder faces. What happens when the next version of Claude makes my product a feature? If you sell a tool, you're in a race against the model. Every upgrade from an AI lab threatens to commoditize your product. But if you sell the work, every improvement in the model makes your service faster, cheaper, and harder to compete with. 

Today's judgment becomes tomorrow's intelligence. This is the subtle but critical point in Bek's framework. Roetzer illustrated this with his Tesla's full self-driving: "When I'm driving my Tesla home from the airport Thursday night, it nails a pothole. Didn't see it. That's an instance where the human judgment has to come in. But at some point, maybe the car gets better than me at judging it."

Outsourced work is the wedge. Bek identifies three reasons outsourced work is the ideal starting point for autopilot companies:

1. The company already accepted the work can be done externally.

2. There's an existing budget line that can be substituted.

3. The buyer is already purchasing an outcome rather than a tool.

Roetzer added a critical fourth reason: "You don't have to fire anybody. So if you're already outsourcing it, then the best thing you can do is get rid of that. Now, that's not great if you are the company providing the outsource services. But it at least buys you time to not have to lay a bunch of people off."

The verticals are massive. Bek mapped the professional services industries most ripe for autopilot takeover: insurance brokerage, accounting and auditing, healthcare revenue cycle (i.e. billing and medical coding), recruitment, management consulting, and legal transaction work. Each one represents a category where intelligence work dominates and AI can sell the completed outcome.

Accounting as most needy and most vulnerable. The profession has lost roughly 340,000 accountants over the past five years, and 75% of CPAs are nearing retirement.

"That just accelerates someone building AI to do that job. And as soon as you fill the gap of those 300,000 to 400,000, you've now automated the need for anybody else."

— Paul Roetzer, founder and CEO of SmarterX

The talent shortage creates the demand for AI, and the AI that fills the shortage eliminates the need for the talent entirely.

SmarterX Take

Start with what you outsource. If your organization already pays external vendors for completed work, whether that's bookkeeping, content production, legal review, or recruiting, that's your immediate starting point. The budget exists, the outcome is defined, and no one on your team loses their job.

The copilot-to-autopilot transition is happening at different speeds. Software engineering is nearly halfway there (49.7% agent deployment). Highly standardized professional services such as insurance brokerage and accounting are next. Judgment-heavy fields including management consulting will take longer. But "longer" with AI might only mean a few years.

The shift from copilots to autopilots isn't a prediction about what might happen. It's a description of what's already happening, backed by the money flows and deployment data to prove it.

What to Watch

• Which verticals tip first. Insurance brokerage and accounting have the clearest path because of their standardized work, fragmented markets, and acute talent shortages. Watch for the first AI-native company to hit $100M in revenue selling completed work in either category.
• The finance job openings collapse as a leading indicator. A 75% drop from peak to levels below the 2001 recession is not normal cyclical movement. If other professional services categories follow the same curve, Bek's thesis is playing out faster than even he projected.
• Whether "copilot" branding ages poorly. Microsoft built an entire product strategy around the copilot metaphor. If the market moves to autopilots that sell work, not tools, the copilot framing may become a liability.

Resources

• Sequoia Capital: Services — The New Software
• Julien Bek on X

Heard on Episode 201 of The Artificial Intelligence Show: This story was explored in depth on the podcast with SmarterX and Marketing AI Institute founder and CEO Paul Roetzer and Mike Kaput. To hear more about Julien Bek's essay and its implications, listen to the full episode. Listen →