video1.0<iframe src="https://www.loom.com/embed/2b11a2003b654ecc989a54ecc178c10a" frameborder="0" width="1152" height="864" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>8641152Loomhttps://www.loom.com8641152https://cdn.loom.com/sessions/thumbnails/2b11a2003b654ecc989a54ecc178c10a-b0a3c01f13213cd2.gif204.134This Loom describes an AI Engineer Hakoton submission for deterministic, auditable property value estimates using comparable sales analysis. Bertrand explains that the LLM layer only rewrites the final result into a memo, while the system handles comp search, scoring, valuation, confidence, and risk flags first. He demonstrates the app using an Edmonton detached home scenario, where synthetic data is loaded, comps are found and ranked with a B-estimate value plus confidence, and reasoning appears in a top comps table. He also notes the modular Python structure for validation, data loading, filtering, and scoring, and that synthetic data was chosen because clean public transaction price data is hard to access.