The community built for founders

Enquanto um projeto ainda não deu certo, as pessoas vão nos chamar de teimosos ou loucos por continuar tentando. Mas quando a ideia finalmente funciona depois de anos de esforço, essa mesma opinião muda completamente. Aquilo que antes chamavam de loucura passa a ser visto como genialidade ou visão de mercado. No fundo, a ideia não mudou e o processo continuou sendo o mesmo; o mundo é que só valida o esforço depois que ele traz resultados.

Not paused it. Not pivoted it. I was genuinely considering deleting the whole thing and moving on. The funny part is that nothing dramatic happened. I didn't lose a huge customer or run out of money. I was just tired. I had been working on it for months, checking analytics every morning, trying different things to get users, posting here and there, fixing bugs, adding features. And the numbers were still... not moving much. One evening I opened the dashboard and saw basically the same numbers I had seen the week before. I remember thinking, “What exactly am I doing?” I even started looking at other ideas. Suddenly every random idea on Twitter seemed better than the thing I had already spent months building. I didn't delete it in the end. I left it alone for a few days instead. When I came back, I realized I wasn't actually tired of the product. I was tired of expecting something to happen every single week. That's probably the part nobody tells you about building from zero. There are long periods where you're working, but nothing visible seems to be happening. I'm still nowhere near where I want the startup to be, so this isn't one of those “I almost quit and then hit $100K MRR” stories. I'm still in the middle of it. I just didn't expect the hardest part to be staying interested when the numbers aren't giving you much reason to be excited.

What do these new AI model releases actually change? I’ve been using Claude for coding and ChatGPT mostly for writing, so I understand the difference when I’m actually using them. But whenever Anthropic, Google or OpenAI announces a new model, I sometimes wonder what I’m supposed to notice. They’ll say things like “new generation,” “better reasoning,” “better coding,” “better agentic capabilities” and everyone gets excited. But what does that actually mean for someone building a startup? Is the improvement mainly in coding? Does Claude become better at understanding a large codebase? Does ChatGPT become noticeably better at writing? Does Gemini become better at research and working with huge amounts of information? Or are these upgrades also about things happening behind the scenes, like using tools, browsing the web, handling longer tasks, understanding images/video, following complicated instructions or making fewer mistakes? Because I can use an older model today and then try the newer one tomorrow and sometimes think, “Okay... what exactly changed?” I know benchmarks can show huge improvements, but I’m more interested in the practical side, If I’m using AI to build a SaaS, what should I actually expect from a model upgrade? Does it mean the AI can now build things it genuinely couldn't build before? Or does it mostly mean the same tasks become a little more reliable, faster and require less correcting? And for founders using AI every day, how do you personally notice a major model upgrade? I’m asking because I feel like we’re getting so many new versions now that “new model” doesn't necessarily mean “completely different experience” anymore. Maybe I'm misunderstanding what a model upgrade is actually supposed to represent.

This guy makes $13K/month from a game he built himself and honestly, the part that surprised me wasn't even the money. I just watched an interview with Ken, who left Meta after years in big tech and started building his own stuff. He ended up making a mobile text-based RPG called Harpagia. The game now does around $13K/month and reportedly made about $155K in its first year. But what caught my attention was how he got there. No big marketing budget. No huge team. He was basically posting on Reddit, talking to people, taking criticism, improving the game and doing it over and over again. At one point he said some of the posts that helped him the most weren't even universally positive. People were actually discussing and criticizing the game. That part is easy to overlook. Everyone loves the screenshot of "$13K/month." Nobody wants to talk about making the same small improvements for months when hardly anyone is watching. He also used Cursor and ChatGPT to help build the game after not writing much code for several years. So you have a former Meta engineer, building a weirdly specific game, using AI to speed himself up, and growing it without paid marketing. And the numbers are pretty wild. A few thousand active users. Around 6% conversion. About $500/month in operating costs. That is a very different picture from needing hundreds of thousands of users to build a real business. The biggest thing I took from the interview wasn't even the AI part. It was the consistency. He launched before everything was perfect, listened to players, kept improving the game and stayed with the same thing long enough for it to compound. I've been seeing so many founders looking for the next growth trick lately. Sometimes I think the boring answer is still the one that works. Build something specific. Put it in front of people. Listen. Improve it. Repeat it for a very long time. The full interview is worth watching if you're building solo: