Finding real estate today relies on the filtered search approach: select specifications and value ranges from a menu which filters out properties failing to satisfy them. This approach burdens the user with knowing which characteristics and values are important for their lifestyle, and fails to match users to properties that satisfy hard-to-quantify and hard-to-categorize preferences. AI-based recommender systems can alleviate these difficulties by returning properties based on more intuitive selection methods such as fuzzy feature selection and similarity to other properties of interest. But the unique aspects of real estate require a novel approach to the recommended engine. Using user responses to a dynamic sequence of lifestyle questions we can assign weights to appropriate property characteristics and use them to score properties based on supporting that lifestyle. Such a system can complement traditional filtered search with chat-based search, virtual agents, and agent-assist tools.