Which Industries Will Be "Solved" and Where Will Jevons Paradox Kick In?

There has been a lot of talk about replacing jobs with LLMs and agents. No doubt, many job functions are going to be automated. At TIDY, we are actively attempting to automate almost all of the work of property management—and the word "almost" matters a lot.

Others have declared the demise of software engineering. Others invoke Jevons paradox: as something becomes cheaper and easier, we consume more of it, not less. So which is it? Are we watching huge industries get completely displaced, or are we about to discover how elastic demand really is?

We won't try to predict the super-distant future. But over the next 20 years, professions will be shaped by a simple question: how much does automating some functions actually increase total demand? That ratio—call it the Jevons multiplier—combined with Baumol's cost disease, explains almost everything about who grows, who shrinks, and who gets "solved."

Why we're thinking about this

This isn't a theoretical exercise for us. We spent 13 years building TIDY, and in the process we've automated essentially every sub-task of property management: scheduling, messaging pros, triaging maintenance, inspecting turnovers, handling guest issues, coordinating vendors, running the digital twin of each property. The job a property manager used to do by hand is already, as of now, automatable end-to-end. Today, TIDY works for hundreds of property owners and managers every month with zero human intervention on our side.

We're not predicting disruption for property managers. We are watching it happen in real time. The diffusion phase has started—most of the industry hasn't adopted it yet, but the technology is here and the economics are unarguable.

Which means we spend a lot of time thinking about the ramifications. What happens to the people? What happens to the market? Where does the demand actually go when the floor cost falls to near zero? All of this is new enough that we're still wrapping our heads around it. This post is us thinking out loud.

The Jevons multiplier

Jevons noticed that more efficient steam engines led to more coal consumption, not less. Make a resource cheaper and we find a thousand new uses for it. But this only works where latent demand exists. If you already have all the accounting you need, making accounting 10x cheaper doesn't make you buy 10x more of it.

The multiplier depends on three things:

• Unmet demand: How many people or businesses wanted this but couldn't afford it?
• Elastic use cases: Once the price drops, do new uses open up that didn't exist before?
• Bottleneck shifts: When this step is automated, does the next bottleneck create new demand—or a dead end?

Chesky's 11-star framework, and why it matters more now

Brian Chesky has a famous thought experiment about what a 6-star, 7-star, all the way up to 11-star Airbnb stay would look like. A 5-star stay is smooth and predictable. A 7-star stay has a limo waiting at the airport and a surfboard in the closet because the host knew you surfed. An 11-star stay has Elon Musk flying you to space. The point isn't to actually ship an 11-star experience—it's that once you map out what truly exceptional looks like, you can work backwards and figure out how far up the scale you can actually reach.

This framework matters more now, not less. The normal, predictable parts of service—the 5-star baseline—are becoming cheap and perfectly executed by AI. Which means firms will fundamentally compete on two things: cost and quality. At the cheap end, AI wins on both. Having humans do tasks a machine can do will steadily become a disadvantage.

So if humans want to add value, they'll add it at the extremes: the hand-done, the physical, the unexpected. The 7- through 11-star stuff. Greeting at the door. The parade. The trip to space. The things that are so clearly human and so clearly specific to you that no machine could have produced them.

This is also where Baumol's cost disease kicks in. Baumol noticed that in sectors where productivity is hard to raise—live orchestras, bedside nursing, one-on-one teaching—wages still rise, because workers have other, more productive options. The same dynamic is about to apply to premium human service. As AI makes the normal end of service radically cheaper, the human-delivered premium end gets more expensive in relative terms. A 5-star baseline becomes effectively free; a human-staged, human-run, hand-delivered 11-star experience becomes a luxury good with luxury pricing. The top property manager, contractor, or concierge doesn't disappear. They get paid dramatically more for work that's harder to copy.

That's the general pattern we expect across most service industries: AI handles the normal; humans handle the extreme and the physical, at a premium. The middle—humans doing what AI can already do—evaporates.

Software: nearly infinite latent demand

Software is the canonical Jevons case right now. Even while headlines declare software engineering dead, jobs are generally up. New games, custom apps, internal tools, weird one-off agents—people want all of it. A beginner now plans and reviews more than they type, but output is up.

The reason is simple: almost every business and person has software they wishthey had but never built because it was too expensive. That backlog is enormous. As coding gets cheaper, the backlog starts getting drawn down, and new ideas arrive faster than the old ones can be shipped.

Accounting: bounded demand, headed for "solved"

Is there infinite demand for accounting? LLMs plus software are likely to have solved most accounting needs within a few years. Will people demand more? Maybe—more audits, more creative reporting, more real-time financial modeling. But a bakery does not want more accounting beyond a point. They want enough accounting, correctly, cheaply.

When demand is bounded, the industry competes on price. Expect exotic business models: free accounting bundled with your bank, your insurance provider, or your payroll processor. The surviving human accountants will either specialize in the hard, weird edges (complex tax, forensic, M&A) or disappear into software companies.

Entertainment: bounded by time, not money

Entertainment feels unbounded—people always want more. But it's constrained by a hard wall: hours in a day. Netflix has said their biggest competitor is other forms of time: social networks, games, sleep. Two- and three-screen viewing (phone plus TV) roughly doubled attention, but you can't scale that much further.

Cheaper production means more varieties of entertainment, not more entertainment consumed. Niche wins. The mass middle gets hollowed out. Brain-computer interfaces might eventually expand the ceiling on attention, but that's a separate bet.

What agents and augmented humans will want

Another bet is on the things agents will want. If an agent can spin up a niche business the moment it sees an unmet need, the plumbing around that—domains, payments, compliance, logistics, small-batch manufacturing, delivery—sees demand explode. Anything where the marginal return is higher than the cost of compute gets bid up.

People will fluidly create micro-businesses the way we create throwaway side projects today. The infrastructure layer—cloud, payments, identity, logistics—gets a Jevons boost. The work inside those businesses is largely automated.

Property management: expansion, not extinction

For property managers, we expect growth to come from market expansion, not from doing more per existing customer. Most long-term rental owners don't use a property manager today because of the expense. Almost no homeowners use one for their own home. If it were good and cheap, many would.

There are about 350,000 property managers in the U.S. today, and most do a daily grind of coordination work. That grind is exactly what agents are good at right now. So what happens to those 350,000 people?

• Specialization: staging, setup, renovation coordination, and "getting a property ready to perform"—work that's hard for agents to do physically.
• Extreme service: hand-done, high-touch experiences that AI can't deliver. For short-term rentals: greeting at the door, curated neighborhood intros, hand-picked groceries. For long-term rentals: in-person tenant placement, walking an owner through a complex renovation decision, resolving a sensitive owner/tenant dispute face-to-face.
• Upgrades and renovations: cheaper ongoing management means more budget freed up for capital improvements. Construction and renovation demand goes up.
• Onboarding and trust: humans will still help owners set preferences, feel comfortable, and make the jump. But not run the daily loop.

The bigger prize: cleaning and maintenance demand itself

Here's the part that gets missed when people argue about whether property management "gets solved." Property management sits next to a much larger market: cleaning and maintenance. And cleaning demand is nowhere near saturated.

Think about how often your home actually gets cleaned versus how often you wishit got cleaned. Think about the baseboards, the windows, the garage, the car, the gutters, the grout, the fridge, the mattress, the air ducts, the deck. For most households, cleaning is heavily rationed on price and coordination friction—not on desire. Both of those frictions fall dramatically when an AI layer coordinates pros seamlessly.

Start with penetration. Today, only about 10% of US households pay for a professional cleaner. Raise that to 50% and the market roughly 5x's on penetration alone. Now layer on frequency: a weekly clean becomes a twice-weekly baseline. Seasonal deep cleans become monthly. "Before guests arrive" becomes a one-tap flow. Car details, post-contractor cleanups, post-party resets, pre-listing refreshes—all become routine instead of rare. We don't think cleaning demand is infinite. But people could plausibly consume 10x more cleaning services than they do today, and maybe 100x once it's as easy as asking an assistant.

This is where the real Jevons multiplier lives for our market. Property management is the wedge. The larger prize is that cleaning and maintenance demand itself expands by an order of magnitude or more once AI removes the coordination cost.

How to tell if your industry is about to be "solved"

A quick checklist:

• Is there a huge population of potential customers who don't buy today because it's too expensive? If yes, you'll grow.
• Does cheaper unlock new use cases that didn't exist before? If yes, you'll grow.
• Is the work a closed loop—clear inputs, clear outputs, limited creativity? If yes, prepare to be bundled for free into something else.
• Is the bottleneck physical presence, trust, taste, or accountability? If yes, humans stay—but fewer of them, at a premium.

Where TIDY fits

At TIDY, we're not betting that property management disappears. We're betting it gets massively larger—because most properties that should have a manager don't have one today. AI does the daily work; humans do the setup, the edge cases, the hospitality, and the upgrades.

The bigger bet underneath that: property management is a wedge into the much larger cleaning and maintenance market, where we expect demand itself to grow by an order of magnitude or more as AI removes the coordination cost.

Take care of your properties like it's 2026, not 1996. See how TIDY works.