How Julius Survived OpenAI's Shutdown and Built 2 Million Users

"I quit my job and then it took a year and a half for me to launch Julius after that. When I went all in, I was nervous, to be honest, I was scared, but it's being a little scared is good. I think fear is good."

This is Rahul Sonwalkar, founder and CEO of Julius AI, reflecting on the moment he burned all bridges to build what would become one of the fastest-growing AI data analysis tools. In just 18 months, Julius has attracted 2 million users who have created over 10 million data visualizations, with the platform now writing 4 million lines of analysis code daily—more than an army of data scientists could produce.

In this candid interview, Rahul reveals why focused AI tools will beat general-purpose giants, how he survived an existential crisis when OpenAI shut down their biggest user acquisition channel overnight, and the counterintuitive strategy of launching products "when they're barely working."

Rahul: ChatGPT is like an all-in-one tool, making videos, making images, doing internet search, writing essays, and it's just an all-in-one tool. And it's not focused on data analysis, so the experience is much worse.

A lot of our users, they start with using ChatGPT and they realize that to analyze any kind of real data to get any kind of meaningful insights, ChatGPT falls short. The quality of insights you get aren't that deep. It cannot handle any kind of real data. The charts don't look good. You can't collaborate with your colleagues on the analysis.

So all these problems people run into after using it for a day for analysis and they all go to Google and they look up AI data analysis. And the number one search result on Google for that is Julius.

It's kind of like, think about hiring humans. Would you want to hire one human that can do everything? You can't, and that will help you get decent results. But then there comes a point where you want someone with deep expertise, someone who's really, really good at one function, like a marketer, finance person, an engineer. I think focused agents for a task have much more competitive advantage than like a general purpose agent.

Rahul: You know, there's just like conventional meme, like Google just killed your startup, OpenAI just killed your startup, XYZ just killed this all startup. I think all that is very overblown. As long as your users don't care about that stuff, it shouldn't matter.

If you're building something that solves the problem for people, if you're doing that good job at that, better than everyone else, that's all that matters.

Rahul: So this is back in college when I got into the hackathon scene. I would go to these hackathons. The big problem I noticed is you have 48 hours to build and launch an idea. A lot of the time people would spend on setting up a backend service, setting up their database, and I thought that was a waste of time.

So Waterview was this managed service, you would get out of box like a backend server or database that just kind of worked. It got a lot of users, and I would go to hackathons, give it to people, and they would try it, it would save them time.

But the problem was none of the people that build their hackathon projects, their weekend projects on Waterview can continue to work on those projects after that weekend, you know, it was like a weekend thing.

The big lesson there was you can solve the right problem, you can solve a pain point for people, but if they don't have that pain point daily or weekly, they're not going to retain. They're not going to come back to the product. You have to solve a problem people have, but if they have it only once a year or once every 6 months or once a month, they're not going to retain to your product. That's the reason why we failed, and that was a very valuable lesson.

Rahul: I worked as an engineer at Uber and Facebook. When you're at a big company and you have an idea, one no can kill that idea. You get a yes from your manager, you have to get a yes from your manager's manager, design manager, product manager. You have to get all these yeses and even one no can kill your idea. So innovation doesn't really happen in big companies.

And then at a startup, it's complete opposite. All you need is one yes. Looking for customers, you know, you can get 50 nos, but one yes, your first customer, that's all that matters. Instead of one no killing your idea, one yes can really make your company work.

Rahul: So one of the problems I was trying to solve at Uber is people book Uber rides usually for things that they can't drive to, like an airport or from an airport, or when they're out, going out at night to bars or restaurants. That sporadic usage is not good for Uber.

One of the things I want to solve for is how could we get people to use Uber on a daily and weekly basis. So I'm going to launch this commuter product at Uber. How can we help people use Uber's offerings, Uber Transit, Uber X, or mobility like bikes and scooters. All that as a part of a commuter package that companies could offer their employees.

So in order to pitch the idea, I was pushing it really, really hard. And, you know, I got buy-in from a lot of cross-functional management. I couldn't get buy-in from our engineering leader. And I couldn't get buy-in from the engineering leader and that crushed the idea.

But at the same time, I was building apps and different ideas on the side on the weekends. And that motivated me to just quit my job and spend a year exploring my ideas full time.

Rahul: Every failure is good. Failing is good because you know what's not working. I talked to founders and they're too scared to launch, too scared to tell people what the idea is. They're too scared to put the product out there. I'm the opposite.

You know, at Julius, we launch things when they're barely working. Because you want to get early feedback from our users, from our customers. We want them to try it out. We want to learn from our users and customers on what they actually want and how the things should work.

Rahul: So I'll give you one of the examples. We were building in data and AI and insights. And we thought, you know, one of the group of people that want insights from data, but don't have the expertise to go to data science on their own is sports fans.

We built this thing called NBA GPT. It's called Hoops GPT to help you query NBA data, play by play data, just asking simple questions. We have to build the UI, the interface, figure out how this text to querying engine is gonna work, and ship the whole thing in like 2 weeks. Because if we were too late, the NBA season would be over. There was no point to wait a year to try the idea again.

And it turns out like sports fans aren't that savvy about data. Turns out it's only the people that want to do betting. They really are and they want to serve. So, we learn very quickly, know whether it's something people want or don't want. We don't want to build something that no one will ever want to use.

I think failing is important. Failing fast is really, really important. Paul Graham talks about this in his essays all the time. They like failing fast is super important.

Rahul: This is months after launching Julius, we had about 10,000 to 15,000 users, and most of our users at that point were actually coming from ChatGPT. ChatGPT had a plugin store where users would discover Julius, and they would come to Julius, stay with Julius and use Julius over and over. And it's going great. It's going great. We're growing overnight.

OpenAI announces that the plugin store is going to get shut down. And our biggest source of users disappeared overnight. So, we had to scramble and figure out how we're gonna get more users and where we're gonna get users from. So that was an existential moment and that is what really got us to move fast and figure out, OK, how do we get our users to become champions of Julius.

Then we realized that when people analyze data, they want to tell their colleagues about it. When you get an insight, you don't keep it to yourself. You want to tell your colleagues, you want to tell your team, and so we build sharing into the product.

Rahul: If you solve a problem that people have, they will come to you. They'll come to you and they will use your product and they'll tell their friends about it. So word of mouth is a free way to grow your product.

Rahul: Nothing. I think, you know, all the things I would consider as like missteps, things I could have avoided, I think those are valuable lessons. You know, all the features we tried and didn't work out, all that is really valuable lesson that's helpful data.

You can, if you have only data from successes, then that's very skewed. You don't know what doesn't work. So it's really important to try a lot of things. Things that don't work are also important because now you know what doesn't work.

Honestly, like I would not do anything differently.