Anyone Can Build an App Now, But What Should You Build? Here’s What Y Combinator Recommends

The AI revolution has sparked a renaissance in app creation, empowering both founders and individuals to build applications faster and more efficiently than ever before. Even I'm getting in on the action..

With AI-powered coding tools, complex development tasks are now accessible to those without traditional programming expertise.

But with so many possibilities, where should you begin? The key is to focus on solving real problems—leveraging AI not as a gimmick, but as a tool to create meaningful, impactful applications.

Most people trying to start AI companies get it backwards. They begin with the idea that they want to build something with AI and then try to find a problem that fits. That almost never works. The best AI startups, like the best startups in general, don’t start with a technology—they start with a problem.

Stop Thinking About "AI Ideas"

The worst way to come up with a good AI startup idea is to sit down and think, "What AI startup should I build?" This leads to generic, overplayed ideas like "ChatGPT for X" or "an AI-powered version of Y."

The problem is that these ideas don’t come from deep insights about real problems.

They come from trying to surf the latest trend.

Instead, the right way to find AI startup ideas is to look for inefficiencies in the world. Specifically, look at where people are doing expensive, manual work that AI could replace or augment.

A simple rule: if something is still being done by hand but could be automated, it’s a potential AI startup.

There are obvious places to look: industries that are still stuck in the 1990s, where Excel is the core operating system. Sectors where expensive consultants are being paid too much to do work that could be turned into software.

Tasks that take hours but should take minutes. The best AI startups won’t just replace human effort; they’ll eliminate entire categories of inefficiency.

Key Takeaways:

• Real World Experience: Get a bona fide job in an industry you want to overhaul. Experiencing problems firsthand can provide insights technology alone can't uncover.
• Consultant Conundrum: Where are companies bleeding money on pricey consultants? That's an opportunity waiting for AI intervention.
• Excel Overload: If a company is overly dependent on Excel for critical operations, you’ve likely stumbled upon a sine qua non for automation and innovation.

Go Deep, Not Wide

One of the biggest mistakes founders make is trying to be too broad.

The best AI startups don’t spread themselves thin trying to build something for everyone. They go deep into one problem and solve it completely.

How do you find these problems?

The best way is to get as close to the work as possible. Shadow professionals for a day and see where they struggle. Work in an industry yourself and pay attention to the things that feel broken. Often, the most valuable insights come from experiencing the pain directly.

This is where domain knowledge becomes an advantage.

If you've spent time in an industry, you’ll see problems that outsiders don’t. This is why some of the best AI startups come from founders who worked in seemingly "boring" fields. They lived the inefficiencies and knew exactly where automation would have the biggest impact.

Key Takeaways:

• Shadowing: Spend a day in the life of a professional in the field you're targeting. Firsthand exposure often reveals latent pain points.
• Live the Problem: Find ways to temporarily integrate yourself into the industry you’re aiming to transform. Empathy is your ally here.
• Unspoken Pain Points: It’s the problems whispered about in back rooms, not those tweeted about, that hold the most promise.

Find Your Unfair Advantage - Your Secret Weapons

Most great AI startups don’t succeed because of the idea—they succeed because the founders had an unfair advantage. This can come in different forms:

• Past industry experience – If you've worked in an industry, you have insider knowledge that most AI founders don't.
• Unique data access – AI is useless without good data. If you can get access to a dataset that others can’t, that’s a huge edge.
• An outsider’s perspective – Sometimes, the best founders aren’t industry veterans. They’re outsiders who notice inefficiencies that people inside the industry have learned to accept.
• Overlooked internet communities – Some of the best AI ideas come from niche communities online. These are places where real users are complaining about problems in ways that never make it into tech media.

Great AI startups aren’t just about "being smart"—they’re about seeing something others miss. The best way to do that is to be in a unique position where you have access to knowledge, data, or insights that others don’t.

Key Takeaways:

• Past Experiences: Reflect on past internships or jobs, even the ones you’d rather forget. They might hold the key to a critical insight.
• Outsider Perspective: Your unique view as both a newcomer and an insider could help you spot opportunities others overlook.
• Data Digging: Internet community forums and obscure data sources are treasure troves of under-utilized insights.

Look for the Right Signals - Finding the Hidden Startup Gems

The best AI startup ideas tend to have certain characteristics. If you’re looking for problems that are worth solving with AI, ask yourself:

• Does this take hours but should take minutes? AI is best at compressing time—tasks that used to take a human hours can often be done in seconds.
• Is this problem costing companies $100K+ per year? The best AI startups don’t just save people time; they save them serious money.
• Is this industry still using manual data entry? AI thrives in places where people are still inputting data by hand.
• Does this field feel "stuck in 1995"? If an industry still operates the way it did 30 years ago, it’s probably full of opportunities for AI automation.

These signals don’t guarantee a good AI startup idea, but they make it much more likely that you’re on the right track.

Key Takeaways:

• Time vs. Task: Spot tasks that irrationally consume time—they’re strong candidates for optimization.
• Monetary Penalties: Issues costing companies a pretty penny ($100k+) annually are begging for a cost-effective fix.
• Outdated Operations: Industries straddling the line between vintage and archaic often hold untapped potential.

What Actually Works

The AI startup graveyard is full of companies that built impressive technology but never found real users. The startups that succeed tend to follow a few key principles:

1. Partner with industry insiders early – If you don’t have domain knowledge, find someone who does.
2. Start with a small, specific user group – The best AI startups don’t try to be everything to everyone. They start by solving a very focused problem for a very specific set of users.
3. Focus on the problem, not the technology – AI is just a tool. The real goal is to solve a problem that people care about.
4. Build for revenue, not just users – Unlike consumer startups, AI startups often make money from the start. If your idea doesn’t create clear business value, it probably won’t survive.

The Good News

The AI revolution is changing the math of startups. A decade ago, many ideas were considered "too small" to be worth building. But today, AI allows tiny teams to automate entire industries, making previously unscalable businesses viable.

You also don’t need massive teams anymore. A small, focused group of engineers can build something that disrupts an entire field. Many of the best AI startups reach product-market fit within 6-12 months.

And most importantly, the best opportunities aren’t in the flashy, obvious spaces. They’re hiding in boring industries—the ones where work is still painfully manual and inefficient.

The lesson here is simple. Don’t chase AI ideas. Chase problems.

The best AI startups don’t start with AI. They start with an inefficiency in the world—and then use AI as the tool to fix it.