{"type":"video","version":"1.0","html":"<iframe src=\"https://www.loom.com/embed/54a71c8bd1664dfd8796211b09ad4797\" frameborder=\"0\" width=\"1280\" height=\"960\" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>","height":960,"width":1280,"provider_name":"Loom","provider_url":"https://www.loom.com","thumbnail_height":960,"thumbnail_width":1280,"thumbnail_url":"https://cdn.loom.com/sessions/thumbnails/54a71c8bd1664dfd8796211b09ad4797-3f81a781b856f0a3.gif","duration":801.226,"title":"Self Improving AI OS for Freelancers","description":"This Loom demonstrates an AI OS approach for helping a solo consultancy systematically evaluate and apply to freelance opportunities. Andres explains his hub-and-spoke markdown repo that acts as an agent knowledge system with guardrails, scoring rubrics, pricing history, and human approval so proposals are not sent blindly. He then runs a terminal demo pulling from Contra, noting there are currently 282 open jobs, using 1 to 10 scoring and applying checks like defined end-to-end scope, avoiding full time-only roles, and enforcing a work floor. The process flags disqualifiers, allows countering when clarifications make sense, and references that it has led to about $100,000 in contract exposure."}