In this livestream I revealed the winners of the Dynamous x Kiro AI Coding Hackathon—$17K in prizes across 10 winners—and walked through why certain projects stood out. Before I got into the top 10, I shared how insanely competitive it was: there were dozens of submissions that were genuinely impressive, and the hardest part was narrowing from a “top 16-ish” down to the final 10. I also explained the rubric I used with my team and Kiro’s guest judges (Doug and Jay): application quality and real-world value, how well the builder actually used Kiro (agent workflow, features like sub-agents/prompts), documentation/devlogs, innovation, and presentation. I started with honorable mentions and then counted down the top 10, highlighting what made each project click—RAG done right, tools that improve agent safety, niche-but-life-changing workflows, and even a full RTS game build. My biggest takeaway: “cool idea” isn’t enough. The winners paired solid product execution with a disciplined agentic workflow and clear documentation. If you’re trying to build faster and ship smarter with AI coding tools, these projects are basically mini-masterclasses in spec-driven, agent-first engineering—and they’re a preview of the trends I’m seeing heading into 2026.