7 million apps in 8 months. Which threshold has AI broken through?

On March 16, 2026, in YC's Lightcone podcast, Mukund and Madhav Jha, the two founders of Emergent, presented a set of figures: 8 months and over 7 million apps.

70% of these apps were developed by people with no prior programming experience.

What does this mean?

In the past, there was a threshold for software development. Now, this threshold is disappearing.

Where exactly was this threshold? And how was it overcome?

Section 1 | Where was the threshold in the past?

Many people think that software development is difficult because the technology is hard.

But the real difficulty lies in turning the ideas in your head into something that others can understand and execute.

In the past, almost everyone followed the same path: first find a development team, then explain your ideas, and then the other party starts the translation process.

This is where the problem lies.

You're describing a complete scenario, but what the other party hears is a bunch of scattered requirements. Your thoughts are continuous in your head, but they become fragmented in the hands of the other party. Each additional communication leads to more deviations.

In the interview, Mukund, the founder of Emergent, believes that:

"What's more difficult than money is the problem of expression. I clearly know what I want to do, but when I explain it to others, many details are lost."

Even if you're willing to spend money, it's still difficult to clearly and completely describe what you really want.

As a result, the costs start to pile up.

On one hand, there are development costs. In the interview, it was mentioned that for a slightly complex application, outsourcing the development would cost millions, but now it only costs tens of thousands with Emergent. Another example is that they replaced their original project management software with Emergent internally, saving 20,000 - 30,000 yuan in monthly subscription fees.

On the other hand, there is the time cost. Repeated communication and revisions can drag on for months, or even longer.

There's also a more crucial aspect: control.

When software has to be developed by others, you're actually handing over a very important matter. You can't make changes at any time or test immediately. One user said, "In the past, we had to initiate a project and schedule it. Adding a feature would take a long time to wait." In many cases, you're not even sure if what's finally developed is what you originally wanted.

This is why a large number of ideas stay in people's heads. It's not that no one has thought of them, but they can't reach the stage of actual development.

So, what really hinders software entrepreneurship has never been a single line of code, but the combination of these three things:

Unable to clearly express ideas, unable to reduce costs, and having to wait in line for the development team to add features.

This is the real threshold.

Section 2 | How was this threshold overcome?

The threshold mentioned in Section 1 can be summarized in one sentence: There is a process that must be translated by humans between the idea and the result.

Now, this process is starting to be removed.

Emergent didn't originally focus on helping people write software. They initially targeted software testing.

In real - world development, the slowest part has never been writing the first version, but making revisions, more revisions, and continuous revisions. Each time a change is made, it's necessary to confirm whether it works and if there are any errors. In the past, this step had to be done manually.

They later realized something:

If the verification step can be completed automatically, many things will change.

Once you can automatically check whether a function is correct and if a process can run smoothly, then the process of writing code can progress on its own round by round, without having to be slowly completed by humans.

As a result, the entire process starts to change:

Previously, it was: Human writes → Human checks → Human revises → Checks again

Now it becomes: Input idea → Automatically generate → Automatically check → Automatically adjust

The most time - consuming and error - prone part in the middle has been compressed.

More importantly, it's no longer just about creating an interface that looks good.

In the interview, it was mentioned that in the past, many tools could only create prototypes, but now they can be directly launched online.

What does this mean? In the past, you could quickly create a prototype, but getting a version that can actually be used, run, accept orders, and process data would take a long time. Now, these two steps have been combined. After you state your requirements, it not only gives you a page but also sets up the underlying processes, data, and operating methods.

This explains why the cost has decreased.

In the past, outsourcing software development would cost millions because you had to pay the labor costs of an entire team: product managers to understand requirements, designers to draw interfaces, engineers to write code, and testers to find bugs. Each modification required repeating the entire process.

Now, Emergent has built a complete set of infrastructure to automatically complete these steps. You don't need to maintain a team or have repeated communication and revisions, so the cost has dropped from millions to tens of thousands.

Control has also been regained.

There was a detail in the interview: Someone within the company used Emergent to create a project management tool. There was no need to initiate a project or schedule it. You just say you want to add a feature, and it can be used right after the modification.

This is a bigger change. When a person can submit requirements, see the results, and directly use the software without having to repeatedly explain to others, wait, and make revisions, you don't have to rely on others for help.

So, on the surface, this round of change is about AI writing code, but what's really happening at a deeper level is: from "translation" first and then implementation, it has become "direct expression" for implementation.

Section 3 | Once the threshold disappears, the first thing to change is not technology

Once it becomes possible to create software through expression, the change is not just about the tools.

The bigger change is about people.

A figure was mentioned in the interview:

"80% of users have no programming background."

This was almost impossible in the past. For software development, either you had to be able to write code yourself or find someone who could. Now, it's different. People who understand the business can create software without having to find programmers.

For example, a small business owner in equipment installation. He knows best what customers will ask, what configurations they will choose, and at which step they will hesitate. In the past, these details had to be communicated layer by layer to the developers. Now, he can simply describe these processes, and they become the tool itself.

Another example is a lawyer - entrepreneur. He's not a programmer, but he knows better than anyone what lawyers do repetitively every day. So he created a customer management system that perfectly fits his own scenario.

There's an even more extreme example. An entrepreneur combined psychological counseling and sports training to create an interdisciplinary application. She couldn't find a suitable software anywhere, and the quotes from outsourcing companies were too high. Later, she created it on Emergent herself. It was launched a few weeks ago, and now hundreds of users are using it.

Someone even told the founders of Emergent that a company received a $4 million investment based on an application developed on Emergent.

As more and more people start creating their own software, a second change has emerged:

Software is no longer a unified version; everyone creates their own.

In the past, software was designed for many people, so it had to be standardized, universal, and adaptable to different users. Now, it's the other way around. Everyone is creating a version that's just right for themselves.

There's an interesting detail in the interview. There was a tool within the company used to replace the original project management software. The original software didn't fit their work style well, so a test engineer created one himself. There was no need to initiate a project or schedule it. Just say you want to add a feature, and it can be used right after the modification.

A third change has also emerged: The number of things a single person can accomplish is increasing.

In the past, a product required at least a product manager, a designer, and an engineer. Now, these three roles are starting to be combined in one person. The intermediate steps that required division of labor have been compressed.

In the interview, an entrepreneur said, "I'm the only developer in the team, and other members only need to focus on the business."

This trend is becoming more and more obvious: Technology no longer requires a dedicated team; one person can handle it.

So you'll see that the number of employees in a company doesn't increase significantly, but the things they can do are increasing. In some scenarios, one person can even complete the work of a small team in the past.

Once the threshold for software development is removed, whether you can write code is no longer the key.

What's important are two things:

• Do you understand your own business?
• Can you clearly state what you want?

So, which threshold did AI remove?

It's not just the technical threshold, but also the threshold of "turning ideas into reality."

In the past, this threshold kept 7 million ideas in people's heads.

Now, they've all been developed.

Original article links:

https://www.youtube.com/watch?v=8SVocWnDHwE

https://www.ycombinator.com/companies/emergent

https://emergent.sh/

Source: Official media/Online news

This article is from the WeChat official account "AI Deep Researcher", author: AI Deep Researcher, editor: Shen Si. Republished by 36Kr with authorization.