The End of the Average App

There is something striking about the last decade of startups. If you judged solely by the headlines, you would think the era was defined by consumer apps. We obsessed over them because we used them. While the cultural spotlight was on the rare consumer giants like Uber and Airbnb, the real value was being captured by a sector most people consider boring: enterprise software.

If you look at the IPOs, the difference is stark. You have Workday for HR, ServiceNow for IT, Atlassian for project tracking, Zoom for meetings, and Snowflake for data. These aren’t companies that try to be cool. They just solve problems for large businesses. Yet they quietly became multi-billion dollar successes. In fact, enterprise software accounts for nearly 16% of all unicorn startups in 2025; that is almost double the number of consumer internet companies.

Why did the unsexy companies win? It wasn’t because the software was better. If you’ve ever used corporate tools, you know they are rarely delightful.

The reason is the cost of code. For most of the 2010s, building software was expensive. A good engineering team costs millions a year. To justify that expense, you need customers with a high willingness to pay and a standardized set of needs to recoup those costs. Big companies have big budgets, but more importantly, they have similar problems. A sales team is a sales team, whether it’s at a bank or a tech firm. This similarity, and legibility, is why a company like Salesforce can sell the same product to 150,000 different organizations. By selling essentially the same software to thousands of companies, an enterprise SaaS can generate huge recurring revenues to justify its engineering spend. The high fixed cost of code demanded standardization: you had to build for the “average” user in order to make the business math work.

For the same reason, consumer software defaulted to building for the average user too. But consumers are the opposite of enterprise. They are messy, varied, and specific. Few consumer software startups exist specifically for left-handed guitar players or retirees managing chronic pain, first-generation immigrants navigating a new country’s tax system or parents of children with ADHD. Millions of people exist in these markets, but the software business does not. Building these tools was never economical in the past. In the high-cost era of software, a startup couldn’t afford to build completely separate apps for every niche or demographic. Instead, consumer app startups tried to create one product that everyone could use. The result was a graveyard of so-called “prosumer” tools that tried to do everything for everyone and ended up satisfying no one.

Consider Evernote. It aspired to be a digital second brain for every user—from students to chefs to CEOs. To chase this broad appeal, Evernote piled on feature after feature, until the app became a bloated jumble that pleased nobody in particular. Or look at Mint, the personal finance app: it gave every user the same generic pie charts (“maybe cut back on lattes!”) and surface-level budgeting tips. Mint couldn’t afford to actually act as a tailored financial advisor for each user, so it settled for one-size-fits-all advice that eventually felt trivial.

Even Notion, arguably one of the best productivity tools of the decade, found that the only way to support its soaring valuation was to focus on enterprise clients. Notion started off as a darling among individual users, but it ultimately pivoted hard toward the enterprise: today, they’re selling knowledge management systems to entire companies, not a note-taking app to you or me. This wasn’t an accidental shift; Notion saw enterprise adoption grow 350% year-over-year, proving that business customers were the true key to its success.

The pattern was clear: when software development was costly, the big wins went to products that could standardize a complex problem and sell that same solution to thousands of paying businesses. B2B SaaS thrived by catering to the common denominator. Meanwhile, consumer software that tried to apply this same logic to individuals often failed, because individuals aren’t standard.

Social platforms like TikTok, Instagram, YouTube, and Reddit were the exception. They thrived because they solved personalization through a back door. The software interface stayed standardized while the experience was hyper-personalized through algorithmic content curation. You do not need to build different apps for different people if you can serve different content to them. This worked brilliantly for entertainment because entertainment is passive. But you can’t UGC your way to a product that actually does work. The personalization needs to extend beyond what you see to how the app functions. You don’t need a tax app to show you a feed of what other people are doing; you need it to take your specific numbers and apply the correct rules to them.

However, the cost of writing code is now collapsing toward zero. This changes the economics in two ways. First, the traditional SaaS model faces compression. There’s a growing sense in the market that the seat-based software economy is on shaky ground, thanks to AI. In the traditional SaaS business model, revenue grows with the number of human users (“seats”) at a client. The more employees using your software, the more licenses you sell. But what happens when some of those employees are AI agents? Atlassian, once the poster child of the SaaS era, saw its stock plunge nearly 35% over a six-month span in the second half of 2025.

Second, the difficulty of building a product no longer serves as a moat. Competitors can clone features in days when code is cheap. Every profitable need will be flooded with AI-generated alternatives.

However, the same force that threatens enterprise SaaS unlocks the missing apps. You no longer need 150,000 identical customers to justify development when building costs drop 90%. You can build for the left-handed guitarist and the retiree with chronic pain as a viable business. The same underlying AI system can generate a fretboard trainer that mirrors left-handed chord shapes. It can create a pain management tool that adapts to specific conditions and medications. The software offers different interfaces and workflows based on who you are and what you actually need. In the old world, a software company built one product and tried to get a million people to use it. In the new world, a software company can build a system that generates a different, personalized product for each person.

Think about what that means: we can finally serve all those niche and individual needs that were previously too expensive to bother with. There is a huge latent demand that was never fully tapped by the one-size-fits-all apps. Every organization has hundreds of little Excel or Notion trackers, every consumer has personal goals and workflows: many of those could be apps or features if it were cheap enough to deliver them. And now it’s becoming cheap enough.

We will be headed from mass production to mass customization in software. There will be no more average app, because the average user never existed. There will only be my app and your app, each one delivered on-the-fly by a generative system. The successful companies of the future will be the ones that give individuals exactly what they want in a way that old mass-market software never could.

So, we now have infinite supply and infinite demand; this sounds like a utopia. But in a world where AI can generate countless tailored apps, features, and content for every niche, how does any given product or provider get noticed?

If AI-generated apps flood the market, users will be drowning in options. App Stores and web platforms will be saturated with auto-generated solutions for every micro-vertical and persona. Traditional digital marketing tricks like cheaply buying Facebook ads to acquire users won’t be the goldmine they once were. When there are millions of personalized offerings, ad markets get crowded and expensive, and generic messages get tuned out. In short, the arbitrage of buying attention is gone. The algorithms will be overflowing with AI-generated noise, and standing out will be harder than ever.

In this environment, trust is the ultimate differentiator. Users will turn to products from sources they feel are authentic, credible, and aligned with their needs. Brand in this new era is the accumulation of credibility one earns by being genuinely useful and transparent over time. For many next-generation founders, this means you have to build your audience in parallel with building your product. It’s no coincidence that we see developers on X, YouTube, and Substack sharing insights and attracting followers. They’re cultivating trust and a community, so that when they do launch a product, it’s not just one more AI-generated slop in the ocean of noise. In practical terms, media becomes part of the product strategy. Writing, teaching, open-sourcing, engaging with user communities is how you stand out when discovery is the limiting factor. The products themselves may be easy to clone or spin up, but a loyal audience and a reputable brand cannot be copy-pasted.

The last decade belonged to those who could standardize complex software and sell it to corporations. The next decade belongs to those who can personalize solutions for individuals; and guide those individuals to find and believe in their product. The average app is ending, and with it ends the era when success meant capturing the median user. In the new era, success will mean capturing the hearts of users one by one, at scale.