Yasser posted a demo video to his 16 Twitter followers showing a chat interface streaming responses from uploaded PDFs. He kept it under 20 seconds to hit the 'aha moment' fast, used a familiar ChatGPT-style UI, and name-dropped the tools he used (Langchain, Pinecone). Those AI companies had zero customers yet and were desperate to show real usage — they retweeted him, creating a viral loop. The tweet blew up and drove the first wave of signups.

• Record a <20-second demo showing the core feature in action, using a familiar UI pattern people already love
• Name-drop every tool/API you're using in the tweet — new platforms will retweet you for social proof
• Frame the demo as 'I built this' to tap into the build-in-public audience
• Post it even if you have <20 followers — the content quality matters more than follower count

Instead of tweeting 'new feature,' Yasser treated each release like a full product launch — writing it so someone who'd never heard of Chatbase could understand and care. He kept the context fresh for new eyes scrolling past, which meant every launch tweet had viral potential instead of just speaking to existing users.

• Write every feature announcement as if the reader has never heard of your product
• Lead with the benefit and the 'what' before the 'how'
• Avoid insider language like 'shipped X' — make it accessible to cold traffic
• Relaunch the same product multiple times with different angles

Yasser went into book subreddits and created free custom ChatGPT bots for popular books, losing money on API costs. He did the same for influencers like Paul Graham and Naval. The goal wasn't revenue — it was to get his domain opened by thousands of people to build domain authority and expose the landing page. Some clicked through, saw the product, and converted to paid.

• Find niche subreddits related to your product's use case and offer free custom versions
• Target influencers and create free tools based on their content (books, podcasts, blogs)
• Prioritize domain opens and backlinks over immediate revenue in the early days
• Use free tier abuse as a growth hack if your product has viral potential

On days when Yasser wasn't launching something new, he paid large LinkedIn and X accounts to post about Chatbase. One sponsored post on a big LinkedIn page brought in $4,000 in a single day. He structured the posts to look native and benefit-driven, not like ads, so they converted well.

• Identify high-engagement pages in your niche on LinkedIn and X
• Negotiate sponsored posts that look like organic recommendations, not ads
• Run sponsored posts on 'off days' when you don't have a launch to keep momentum
• Track revenue per post to find which pages/platforms convert best

Yasser had built Chatbase before the ChatGPT API existed. When OpenAI launched the API, he tweeted 'I built this in 48 hours' using the new API — even though the product was already done. Because everyone was talking about what you could build with the new API, the tweet went viral again and drove a second wave of signups.

• Build on a platform or API before it's publicly available or hyped
• Time your 'launch' tweet to coincide with the platform's big announcement
• Frame it as 'I just built this with [new thing]' even if you had a head start
• Ride the hype wave of the underlying technology to amplify your own launch

Chatbase grew entirely through product-led growth — a Stripe checkout button, a clean landing page, and one core feature (upload PDF, chat with it). No sales calls, no demos, no complex onboarding. Yasser focused on a bug-free MVP with good UI so users were excited, not frustrated. 10,000+ customers paying $40–$500/mo, all self-serve.

• Launch with one core feature that solves a clear problem, not a feature buffet
• Put a Stripe button on the landing page and let people buy immediately
• Obsess over UI quality and zero bugs — frustrated users don't convert or retain
• Skip the sales team until PLG revenue proves the model works

Instead of locking into one AI provider, Chatbase integrated OpenAI, Anthropic, Google Gemini, Cohere, and Grok. Users get a dropdown to choose which model works best for their use case. This differentiated Chatbase from wrappers, reduced vendor lock-in risk, and let power users optimize for speed, cost, or quality.

• Integrate multiple AI providers so you're not dependent on one API's pricing or limits
• Let users choose the model in your UI — it's a feature and a hedge
• Use Vercel AI SDK or similar to abstract provider differences
• Position this as 'flexibility' and 'future-proof' in your marketing