The story of my fuck up in the area of my strongest expertise: how Stacy tried to create fully automated communities When I landed on the vision of building the distribution system first (before any products) I started vibecoding content automation systems. My initial idea was to find areas of strongest human interest (ideally fast-growing topics where demand for content is still unfilled) and build a content engine that would pull relevant data and feed it into these communities. It seemed easy: collect news and data, have AI summarize the key points, and voila. A self-serving content engine, and I’d only need to handle promo. I was VERY much mistaken. Building the system itself was pretty easy. I gave it my ToV, writing rules, and a format grid – it looked solid and felt not too AI. But I didn’t factor in one thing: people are overloaded with content, and they don’t want to follow humanless content. I didn't know it then. So I launched one of the first automated communities. I spent ~$1K on Telegram Ads tests to estimate acquisition costs and user behavior. I optimized for good cost per subscriber (~$0.8) – but those users didn’t stay. They joined, then muted the channel within a few days. Mute = death. I literally paid for bots. I know that 90% of bloggers are now using AI to write content, and I was doing the same, but now I see how people react to AI content in practice. No matter if it's quite good and is based on my own content algorithms, it's just not human. Now I’m rebuilding the design from zero, putting AI where it actually does its job well: data sourcing, engagement analytics, and projecting content fit into drafts. Let's see how v2 evolves :) I share my story here →