How Yasser Elsaid grew Chatbase to $1M ARR in 117 days

Executive overview

Yasser Elsaid launched an AI chatbot builder to 16 Twitter followers and hit $1M ARR in under four months — with no funding, no audience, and no prior viral following. The growth came from a single well-structured demo tweet, repeated relaunch framing, and scrappy distribution tactics that cost nothing.

Timing an unsexy product to a platform wave — and reframing every feature update as a new launch — is the actual growth lever.

The viral tweet that started everything

• Third tweet Yasser ever posted; it went viral immediately
• Three elements made it work: familiar chat-streaming interface (felt like ChatGPT), a demo reaching the "aha moment" in under 20 seconds, and naming the tools used (Langchain, Pinecone)
• Naming new AI tools got those companies to reshare — they needed proof their platforms were being used
• This created a viral loop: their audiences amplified the post

Relaunching at every iteration

• Each new feature was framed as a fresh launch, written for someone with zero prior context
• Treating updates as "feature launches" loses newcomers — framing them as standalone products captures them
• Kept generating distribution without spend between launches via sponsored posts on LinkedIn and other pages
• One LinkedIn sponsored post alone generated $4,000 in a single day

Scrappy growth tactics

• Created free ChatGPT demos for popular books and dropped them in relevant subreddits — losing money intentionally to build domain authority
• Did the same for public figures: Paul Graham, Naval
• Goal: get the domain seen by many people; some would click through and convert

Build in public playbook

• Post every day about what you're building — first 2–3 months are practice, not performance
• Short-form video on X stands out because few people do it
• Be personal: share your story, have a personality, take a genuine stance
• One tweet builds nothing; consistency builds a brand

Product-led growth (PLG)

• No sales team: a good landing page, a clear single core feature, and a "buy now" button
• Core feature at launch: upload a PDF and chat with it
• Ship quality fast — frustrated users don't convert; proud users evangelize
• Balance shipping speed against product quality; don't ship garbage

Building with AI agents

• Read OpenAI's cookbooks and agents SDK docs; read Vercel's AI SDK documentation
• Find a product you can't build yet because the AI isn't capable enough — then build it and wait for the models to catch up
• Yasser built Chatbase before the ChatGPT API launched; when the API dropped, his product was ready and his launch tweet went viral again

Tech stack

• Hosting: Vercel
• AI infrastructure: Vercel AI SDK
• Database: Supabase
• Models: OpenAI, Anthropic, Cohere, Google Gemini, Grok (user-selectable)
• Billing: Stripe
• Affiliates: Dubb
• Docs: Mintlify

Lessons from building

• There is no universal blueprint; best practices shift entirely with context
• Outside advice is worth hearing but rarely directly applicable — trust your own read of your situation
• Yasser's biggest regret: not thinking big enough early; his initial goal was $10K/month in Bali, not a $100M ARR company
• He moved too cautiously because critics called it "just a ChatGPT wrapper" — that got in his head; he'd move faster and bet bigger next time