Redfin built an internal tool, launched around 2016, that uses machine learning to automatically suggest comparable homes and generate pricing analyses — a task that traditionally takes agents several hours to complete manually. The tool handles the bulk of the work while agents make final decisions.

The tool draws on the same machine learning algorithm that powers the Redfin Estimate, supplemented by real-time demand signals such as how many active Redfin buyers are searching for similar homes in that area. Agents can accept, reject, or adjust the suggested comparable homes, and the system learns from those choices over time to refine future suggestions. Engineers observed agents working in person during development to ensure the tool matched real-world workflows. The result reduces a multi-hour manual preparation task to minutes, augmenting agent productivity without replacing the agent's judgment.