IdeaLoop: Daily startup ideas discovered & validated by multi-agent AI

I've been an indie developer for a while, and the hardest part was never writing the code — it was deciding what to build. Every time I wanted to start something new, I'd spend days browsing Twitter, ProductHunt, and HackerNews, dumping links into Notion, then slowly researching markets and competitors. The process was slow, inconsistent, and ultimately driven by gut feeling. Most of what I built had no real validation behind it. So I asked: what if this entire process could run on autopilot? IdeaLoop started as a simple experiment — one AI agent scanning a few sources. It quickly became clear that a single model wasn't enough. Market analysis, technical feasibility, user fit, competitive moats — these require different lenses. That's when I moved to a multi-agent architecture: specialized agents running in parallel, with a master agent synthesizing the results into a scored report. The evolution surprised me. The output quality jumped significantly once agents stopped trying to evaluate everything and focused on what they were designed for. Today IdeaLoop runs daily, automatically. The public archive at idealoop.top/archive is open to anyone — no signup needed. I'd love to hear what you think, especially if you're an indie developer who's struggled with the same "what should I build?" problem.