Yaser posted what was only his third tweet ever to his 16 Twitter followers — a short demo video of Chatbase showing a streaming chat interface (mimicking the familiar ChatGPT UI) that could answer questions from uploaded data. The demo was engineered for speed-to-aha-moment: within 20 seconds viewers could see exactly what it did. Crucially, he named the underlying tools he used — LangChain, Pinecone, and other nascent AI infrastructure providers — which gave those companies an incentive to reshare since they were all new and hungry for real-world usage examples. The retweets from those companies created a viral loop that pushed the tweet far beyond his 16-follower base. After that initial spike, Yaser treated every subsequent feature release as a fresh top-of-funnel launch, framing each tweet so that someone with zero context could understand it without needing prior knowledge of Chatbase. He also rode the ChatGPT API launch news cycle: when OpenAI released their chat completions API, he published a tweet saying "in the last 48 hours I built this" — piggybacking on the massive organic conversation already happening around the new API. This second viral moment brought another wave of new users.
Chatbase
AI chatbot builder — train a ChatGPT on your own data/documents
8 moves, in order
- Launch (Day 1)Twitter viral demo
Posted a 20-second demo video to 16 Twitter followers showing the streaming ChatGPT-like interface. Explicitly named the AI tools used (LangChain, Pinecone) in the tweet, prompting those companies to reshare to their own audiences, creating a viral loop.
Tweet went viral; generated first paying customers within days - Early Growth (first weeks)Product-led growth
Set up a simple Stripe button on the landing page with no sales process. Core feature was 'upload a PDF, chat with it.' Focused on zero bugs and a clean UI so users converted and retained without hand-holding.
2-3 sales per day consistently, leading Yaser to drop out of university - Month 1 2Build in public — Twitter
Framed every new feature release as a standalone launch tweet written for someone with zero prior context — not a 'feature update' but a fresh product story. Posted consistently about what he was building day-to-day.
Sustained viral moments with each release, compounding audience growth - Month 1 4Reddit community seeding
Went into book and influencer subreddits and built free Chatbase chatbots for famous books and public figures (Paul Graham, Naval Ravikant, etc.) at his own cost. Goal was domain authority, brand exposure, and converting curious visitors who landed on the Chatbase landing page.
Increased domain authority and organic discovery; some free users converted to paid - Month 1 4Linkedin sponsored posts
On days without a product launch, paid for sponsored posts with large LinkedIn pages. Structured the posts as direct product showcases.
$4,000 in revenue from a single sponsored LinkedIn post - ChatGPT API launch (Month 2 3)Twitter news cycle hijack
When OpenAI released the ChatGPT chat completions API and the topic was trending, Yaser posted a tweet framed as 'in the last 48 hours I built this' — positioning Chatbase as the premier example of what developers could build with the new API.
Second major viral moment, large new user acquisition spike - 117 DaysProduct-led growth
Maintained pure PLG with no sales team — Stripe button, strong landing page, single core feature. All marketing done externally (social, Reddit, sponsorships); website just needed to convert.
$1M ARR reached in 117 daysMRR $83k - Scale (2.5 years in)Product-led growth
Continued 100% PLG with no funding. Expanded model support (OpenAI, Anthropic, Cohere, Gemini, Groq) and added affiliate program via Dub to incentivize word-of-mouth.
10,000 paying customers, 600,000 registered users, $6.8M ARRMRR $567k Users 600k users
No traditional unfair advantage — he had only 16 Twitter followers. His edge was timing: he built Chatbase before the ChatGPT API existed, so when OpenAI launched it he had a fully working product and a newsworthy hook. He also understood that nascent AI infrastructure companies (LangChain, Pinecone) were desperate for real-world use-case visibility, making them willing amplifiers.
twitter_viral_demo_with_tool_mentions
No channels explicitly called out as failures. He implied early build-in-public posting gets no traction for the first 2-3 months and requires persistence before it compounds. He also notes he was "not aggressive enough" early on and held back growth by thinking too small.