Chatbase: 10 side projects that went nowhere, an 11th built in 2 weeks and tweeted to 16 followers, $10M a year with no co-founder and no investors, and how to build your first AI tool even if you are not technical
Yasser Elsaid built 10 side projects before Chatbase. None of them took off. In early 2023, in his last year of university, he spent two weeks building a simple tool: upload a PDF, then chat with it. He tweeted it to 16 followers. Within a day thousands of people were using it. Three years later Chatbase passed $10 million a year in revenue with no co-founder and no investors. This page breaks down how the 11th project worked when the first 10 did not, how a free toy turned into a business tool, and 5 steps to build your first AI tool even if you are not technical.
Overview
Most people who try to build something online stop after one or two tries. Yasser Elsaid kept going. He interned at Tesla and at Meta. He built side project after side project. Ten of them went nowhere.
The 11th was the smallest of all. It did one thing. You uploaded a PDF and asked it questions, and it answered from that document. He built it in two weeks.
The timing was the whole story. ChatGPT had just come out. Everybody was talking about it. Very few people knew how to make it work on their own stuff. Yasser gave them the simplest possible way to try.
The 10 before
Chatbase came after 10 other side projects. He was a strong engineer with internships at Tesla and Meta on his resume. Skill was not the problem. None of those projects found people who wanted them badly enough.
This is the part most success stories skip. The win you hear about is usually the last of many tries. Each failed project teaches you how to build faster, launch faster and kill an idea faster. By the 11th, Yasser could take an idea from nothing to live in two weeks.
He had expected a return offer from Meta after his internship, and it never came. He was in his final year of university with no job lined up.
The 16-follower launch
He started building in late 2022, right as the GPT-3 API became broadly available. The idea came from what he saw around him: lots of people talking about ChatGPT, very few who knew how to use it on their own work.
The first version was stupid simple. Upload a PDF, get a chatbot that answers questions about it. It was free, with unlimited documents and messages, and there was not even a pricing page.
He launched it in early February 2023 with a tweet. He had 16 followers. It was only the third tweet he had ever sent.
Within 24 hours thousands of people had tried it. His servers crashed twice. AI accounts with big followings shared it, and the traffic kept coming.
After the spike, he kept feeding it. He posted on Product Hunt, Indie Hackers and AI tool directories. On Reddit he posted stories about what he built, not just links to the tool.
From free toy to business tool
Then he made the choice that turned a viral toy into a company. He decided to build a business, not keep a side project. He skipped his final classes to build it and failed two of them.
He listened to what users asked for, and they were not asking for better PDF chat. Businesses wanted a chatbot for their customers. They asked for more people per account, better security, a chatbot that sounded like their brand, and connections to their other tools. They asked to train it on whole websites, videos, spreadsheets and audio, not just one PDF.
So Chatbase became a platform businesses use to build AI customer support agents on their own data. Today those agents can also take actions, like looking up a payment in Stripe or booking a meeting in Cal.com.
The money came fast once it charged. It made a few hundred dollars a month within days of launch, about $10,000 a month by mid March 2023, and about $64,000 a month by May 2023. Five months after launch it was at $1 million a year.
He did this without a sales team or venture money. People who tried the product bought it.
What it cost him
Solo and self-funded sounds like freedom. Yasser describes it differently. "You're pressing the brakes a lot more... because you're just scared that... you might run out of money." And when something breaks, "everything is your fault."
He took about 10 days off in two years.
He would still choose it. "Dictatorships is how you build a good company, a great company." One person deciding means decisions happen fast.
Today Chatbase has a team of 18, 11 of them engineers, and more than 8,000 paying customers. Revenue has passed $10 million a year. The stated goal is $100 million.
Why the 11th worked when the first 10 did not
Three things lined up on project 11 that had not lined up before.
A new wave. ChatGPT had just made millions of people curious about AI. Curiosity was everywhere and easy tools were rare.
One simple job. The first version did one thing anyone could understand in five seconds: chat with your PDF.
He followed the money. The free users were consumers. The paying users were businesses. He rebuilt the product around the businesses.
The first 10 projects were not wasted. They are why he could ship the 11th in two weeks, the moment the wave arrived.
How to build your first AI tool, even if you are not technical
Step 1. Pick one industry
Pick an industry, a niche, a vertical or a type of business. Say med spas.
Step 2. Go where they hang out
Read the YouTube comment sections, the comments on their Instagram accounts, their subreddits and their forums to get a feel for their problems. Then walk into a med spa near you, or cold call or cold email them.
Step 3. Listen to their real problems
Let them talk. Write down what they complain about, in their words.
Step 4. Find a solution, charge for it, then build it
Identify a solution that solves their problem. Tell them the price and ask them to pay before you build it. If they pay, then and only then build your tool with AI like ChatGPT or Claude. In the age of AI, building a tool is no longer only for technical people. Non-technical people can build tools with AI too.
Step 5. Show it to them and improve it
Show it to them and improve it based on their feedback.
Want help with this?
If you are not technical and want help learning AI to build tools or services, feel free to reach out to me at karthik@entrepreneurdb.com.
If you are a business owner and need help implementing AI, you can discuss that with me as well.
I want to learn AI I need AI in my businessThe numbers
| Figure | Number |
|---|---|
| Side projects before Chatbase | 10 |
| Time to build the first version | 2 weeks |
| Launch | Early February 2023 |
| Followers at launch | 16 |
| Monthly revenue, mid March 2023 | About $10,000 |
| Monthly revenue, May 2023 | About $64,000 |
| Annualized revenue, 5 months after launch | $1 million |
| Revenue today | Over $10 million a year |
| Paying customers | 8,000+ (early 2026) |
| Team | 18, including 11 engineers |
| Outside funding | None |
Want 50 more breakdowns like this?
Real businesses with their revenue, tools, costs and how they grew, plus step-by-step guides to start your own. Free.
Get free accessMore breakdowns
Sources
- Omar Waseem, founder interview clip with Yasser Elsaid, X, October 20, 2026
- Supabase, customer story: Chatbase
- Indie Hackers, "How a college student reached $64,000/mo in 6 months by being an AI first mover"; "From viral side project to a $5M/yr B2B AI platform"
- Jetone newsletter, "How one developer built a $6M AI chatbot business in just 2 years"
- Solo Founders newsletter, "$9M ARR, no VC, solo founder"
