video1.0<iframe src="https://www.loom.com/embed/5274384c30e24924bbbb25741876182d" frameborder="0" width="1280" height="960" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>9601280Loomhttps://www.loom.com9601280https://cdn.loom.com/sessions/thumbnails/5274384c30e24924bbbb25741876182d-6cd28ab60e60fcf7.gif793.045This Loom explains an end-to-end AI automation platform for data scraping, AI scoring, and proposal generation. The creator describes a two-tier architecture: a fast Llama3 model that rates incoming data in milliseconds and lets only the highest-scoring matches pass to a 70 billion parameter model on ROC hardware to draft the final proposal, which took about an hour. They also cover the custom Next.js UI styled with Tailwind and enhanced with FramerMotion, including interactive features like one-click copy. Finally, they note real-time updates using a Supabase PostgreSQL database with WebSockets so the dashboard updates instantly with zero refreshes required.