Introduction
how to build AI SaaS to $1M ARR — In this Q&A, I share exactly how I, Yaser, took Chatbase from zero to $1M ARR in 117 days and to $6.8M ARR, using product-led growth, viral launches, and practical AI agent strategies.
Founder Success Story QnA
Tell me about who you are, what you built, and what’s your story.
Okay, so my name is Yaser. I built Chatbase to $7 million in ARR with no funding. I launched it 2 and a half years ago as a solopreneur to my 16 followers on Twitter and it went viral.
How did you come up with the idea for Chatbase?
I followed the bootstrapping movement on X like Peter Levels, all those guys. I was doing internships in FAANG companies but once I got there I realized maybe it’s not for me. I looked at people five, ten years ahead of me and it was not something I was looking up to. When I saw people like Peter and Danny Postma, especially in the AI wave, I thought I can actually do this. I saw demos of how you can build a ChatGPT for your data, but no one was building a product around it. I knew this is a good idea. I didn’t do any validation. I just said if I don’t do this someone else will. I released Chatbase to my 16 followers. It went viral. This was in my final year of university as a side project. When I saw the first sale and then two or three sales every day, I dropped out.
Walk me through the viral tweet that changed everything.
I think this was my third tweet ever. Three things made it work. One, a familiar streaming chat interface — very ChatGPT-like — so taking a glance at the demo was interesting. Two, I minimized time to the aha moment — a simple demo that shows exactly what it can do in the first 20 seconds. Three, I mentioned the tools I was using — like LangChain and Pinecone. Those AI companies were new and wanted to promote real use cases. When I mentioned them, they picked it up, went viral, and that created a viral loop. Most of the initial traction was from that.
Break down your build-in-public playbook. What works?
The easiest thing is to share what you’re doing every day. The first two or three months, if you’re not lucky, it’s just practice; you’re building a brand and a story. One tweet won’t do it. People need to see your face a lot. If you can do videos, post them on X. Not enough people do that, so you stand out. Frame it as: I’m building something, I’m learning a lot, I’m taking you on the journey. Don’t be boring. Be personal. Share your story. Have a personality. Have controversial takes if you truly believe them — that’s how you get people to notice you.
How did Chatbase go from zero to $1M ARR in 117 days?
After the first viral moment, I kept adding more features and framed every new release as a new launch. Some people frame second or third launches as feature launches — new people have no context and skip it. I made each launch make sense to people who don’t know me or Chatbase. I was creative about going viral without spending money. I went into subreddits of books and created a ChatGPT for their book — for free. I was losing money, but the point was getting my domain opened by a lot of people for domain authority and clicks to the landing page. I did the same for influencers like Paul Graham and Naval. Lastly, sponsored posts were big. If I’m not launching something today, I do a sponsored post with a big page. We did one with a LinkedIn page and did $4,000 that day. If structured well, sponsored posts are a very good way to get your product out there.
How big is the business now?
Right now we’re exactly at $6.8 million ARR. It’s growing very fast. It’s only PLG. We just passed 10,000 customers paying between $40 and $500 a month. Registered users are around 600,000.
What is PLG and how did you apply it?
PLG is product-led growth. You have a Stripe button on your website. Do most marketing outside your website. Have a good landing page. People can buy a subscription and put in their credit card. You don’t need a lot of features — just one core feature. In my case: upload a PDF and talk with it. Have a good UI, no bugs, a solid MVP. Find the balance between launching fast and not launching garbage. Build a tool you’re proud of so people are excited to use it and not frustrated.
How should founders learn to build with AI agents?
If you want to build products, go to sources like OpenAI‘s cookbooks and documentation. Read about their agents SDK. Do the same with Vercel‘s AI SDK. Read docs, look at examples, and who’s using them. Learn how those companies tell you to build AI agents. A good idea is find something you can’t build yet where the only bottleneck is the AI model. Start building and bet that in a year GPT-5 (or similar) will make it possible. You’ll have a one-year head start. Something similar happened to me — I started working on Chatbase before the ChatGPT API. When it launched, I tweeted a new product built in 48 hours, and because everyone was talking about the API, it went viral.
What is your tech stack?
Now we use Vercel for hosting and their AI SDK for a lot of AI infrastructure. We use Supabase for the database. We use multiple AI labs — OpenAI, Anthropic, Cohere, Google Gemini, Grok. In Chatbase, you can choose the model from a dropdown. For billing we use Stripe, for affiliates Dub, and for documentation Mintlify.
What surprised you most while building this company?
We assume there’s a blueprint to build a good business. But there are so many ways — how you work, reviews, in-person or remote, culture — it all changes company to company. This is good: you don’t need a lot of experience to start a company. Most advice is good to hear but mostly irrelevant because no one knows your situation like you. Best practices change by situation. People who give good advice will tell you to trust your gut because you know your situation more than them.
If you could go back, what advice would you give your younger self?
I wasn’t thinking big enough. My goal was $10K MRR, move to Bali, indie hacker life. Why not shoot for the stars? Why not build the first $100M ARR bootstrap company? People said it’s a ChatGPT wrapper and that got into my head — it was true at the time. I wasn’t aggressive enough; I was too shy and reserved. If I go back, I’d move faster and put more on the line. Aim higher.
Yaser Method
I focused on fast iteration, public storytelling, and distribution loops that compound — shipping core value quickly, launching repeatedly, and riding platform momentum.
- Ship a simple, valuable MVP fast and polish UI to avoid frustration.
- Frame every feature release as a new launch for new audiences.
- Leverage platform virality by naming/using emerging tools publicly.
Yaser Tools
I use a lean, modern stack to move quickly and support multiple AI models, while keeping billing and docs simple for a PLG motion.
- Vercel + AI SDK for hosting and AI infrastructure velocity.
- Supabase for a simple, scalable database.
- OpenAI, Anthropic, Cohere, Google Gemini, Grok to give users model choice.
- Stripe for frictionless PLG billing.
- Dub for affiliates and Mintlify for documentation.
Key Notes
These are the principles I kept returning to while scaling from zero to millions in ARR.
- Minimize time-to-aha in demos and landing pages.
- Build in public and repeat launches to reach new audiences.
- Create free value to spark distribution (e.g., book and influencer bots).
Get Started in Just 5 Steps
Follow this path to learn how to build AI SaaS to $1M ARR with a practical, fast-moving approach.
- Pick one clear problem and ship a tight MVP with a smooth UI.
- Publish a short demo with an immediate aha moment.
- Name-drop and integrate emerging tools to ride their distribution.
- Treat every feature as a fresh launch for new people.
- Use low-cost channels: Reddit seeding, influencer tie-ins, selective sponsored posts.
Conclusion
If you’re wondering how to build AI SaaS to $1M ARR, focus on fast shipping, compelling demos, repeated public launches, and scrappy distribution. That’s how I scaled Chatbase from zero to millions without funding.