0 to 2M Users in 18 Months: How I Built the Leading AI Data Analyst | Julius AI, Rahul Sonwalkar

Key Concepts

• AI Data Analysis
• Focus in Startups
• Importance of Solving Daily/Weekly Pain Points
• Innovation in Startups vs. Big Companies
• Failing Fast and Learning from Failures
• Word-of-Mouth Growth
• Data Visualization
• User Retention

Launching Julius AI and Initial Fears

Rahul, the founder and CEO of Julius AI, discusses his journey after quitting his job and the year and a half it took to launch Julius. He admits to being nervous and scared but emphasizes that a little fear is beneficial as it keeps you sharp and prevents complacency. He highlights the importance of burning bridges to force oneself to take risks and make things work.

What is Julius AI?

Julius AI is described as an AI data analyst that helps users get insights from their data within seconds and create charts and data visualizations. Since its launch in 2023, users have created over 10 million data visualizations. Julius writes over 4 million lines of analysis code daily, which is more than an army of data scientists could produce. It took about a year and a half to reach 2 million users.

The Importance of Focus for AI Startups

Rahul stresses the importance of focus for AI startups, advising against building general-purpose tools. He argues that startups win because of focus, which is their only advantage. He contrasts Julius with general-purpose tools like ChatGPT, which, while versatile, offer a much worse experience for data analysis. Users often start with ChatGPT but realize its limitations in handling real data, providing deep insights, creating good-looking charts, and enabling collaboration. This leads them to search for AI data analysis tools, where Julius ranks highly.

Focus vs. General Purpose AI

The analogy of hiring humans is used to illustrate the point. While a generalist can provide decent results, there comes a point where specialized expertise is needed, such as a marketer, finance professional, or engineer. Focused AI agents have a competitive advantage over general-purpose agents. Rahul dismisses the fear of larger companies like Google or OpenAI killing startups, stating that as long as the startup solves a problem for users better than anyone else, it doesn't matter.

WaterWe: A Failed Startup and its Lessons

Rahul recounts his experience with WaterWe, a managed service he built during college hackathons to help developers quickly set up backend services and databases. While WaterWe gained traction and saved developers time, the projects built on it were abandoned after the hackathons. The key lesson learned was that solving a pain point is not enough; the pain point must be daily or weekly to ensure user retention. This is why WaterWe failed.

Innovation in Big Companies vs. Startups

Drawing from his experience at Uber and Facebook, Rahul explains that innovation is difficult in big companies because one "no" from any manager can kill an idea. In contrast, startups only need one "yes" to make an idea work. He shares his experience of trying to launch a commuter product at Uber, which was ultimately rejected by the engineering leader. This motivated him to quit his job and explore his ideas full-time.

The Importance of Failing Fast

Rahul emphasizes the importance of failing fast and launching products even when they are barely working to get early feedback from users. He shares an example of NBA GPT (Hoops GPT), a tool built to query NBA data for sports fans. The tool was built and launched in two weeks to coincide with the NBA season. However, they quickly learned that sports fans were not as data-savvy as they thought, and the primary users were those interested in betting, which was not the target audience. This experience reinforced the importance of quickly validating ideas and avoiding building products that no one wants.

The ChatGPT Plugin Store Shutdown

Rahul describes a critical moment when OpenAI announced the shutdown of its plugin store, which was Julius's biggest source of users. This existential threat forced the team to quickly find new ways to acquire users. They realized that people want to share their data analysis insights with colleagues, so they built sharing features into the product. This led to word-of-mouth growth, which Rahul considers a free way to grow a product.

Conclusion: No Regrets

When asked what he would do differently if starting from scratch, Rahul says he would not change anything. He believes that all the missteps and failures were valuable lessons. Data from failures is just as important as data from successes because it helps identify what doesn't work.