How I Built an AI Idea Library Using No-Code and Vibe-Code

As a software creator, I often hit this big snag: figuring out what to actually build next. It’s not that I’m outta ideas — there’s tons floating around — but the real worry is will anyone actually want it once it’s done? To tackle this, I wanted to craft a sort of library filled with AI product ideas that come straight from real user problems. This way, it’s not just random guesses but stuff grounded in actual needs. So, I put together solopreneur.global, a resource for myself and anyone else stuck in that same boat. I broke down the work into two main parts and spent about two weeks total getting everything up and running. First, I set up an n8n workflow to dig into conversations happening in communities like Reddit. This workflow pulls in threads where people talk about pain points related to AI products, filters out the weak signals, and focuses on posts with strong engagement. Then, AI steps in to draft full “idea cards” that include a summary of the original post, a clear problem statement, a solution outline, core features, business model ideas, and some go-to-market strategies. This setup runs daily, so the idea library just keeps growing without me having to lift a finger. Initially, I just dumped all the ideas into a spreadsheet, but that quickly got messy and didn’t scale with the volume. So I flipped the switch to vibe-code a lean web app on top of a Next.js boilerplate. The boilerplate was a solid starting point because it already had built-in services like user management, payments, blogging, and analytics. That meant I only had to plug in the keys and run quick tests instead of building everything from scratch. Plus, the boilerplate gave me a great software architecture, with clear layering, multilingual routing, and support for both desktop and mobile, which made it easier for AI to add new features while keeping everything consistent and maintainable. For the actual build, I used Claude Code. My process with Claude Code goes like this: first, I switch to Plan Mode and get the AI to draft an overall design and implementation plan. After I review and approve it, I have the AI implement the entire module all the way through. This approach turned out to be super efficient, with almost no need for rework or major changes. If you’re curious about Claude Code but haven’t tried it yet, it’s definitely worth checking out. All in all, the whole setup helps me stay focused on building AI products that actually matter. The idea library keeps expanding every day, and the vibe-coded app makes browsing and managing those ideas a breeze. It’s been a game changer in moving from endless brainstorming to concrete builds that solve real problems.