In the ever-evolving landscape of Proptech within the rental space, we embarked on an ambitious project that sought to transform the way users discover and select properties for rent. Our vision was to harness the capabilities of Al and machine learning to develop a recommendation system that would streamline the property selection process. This innovative system aimed to learn and adapt to user preferences, offering a daily curated list of ten new properties, thereby eliminating the need for users to sift through hundreds of listings daily to shortlist the properties they wished to view.

However, our journey with this project unveiled a valuable lesson in the strategic selection of Al initiatives. While our initial enthusiasm was fueled by the promise of Al’s potential, we encountered significant challenges that underscored the complexity of accurately understanding and predicting user preferences. Our machine learning model’s performance fell short of our expectations, primarily because it demanded a level of user commitment and input that proved to be impractical for many users to provide.

This experience served as a poignant reminder of the critical importance of carefully assessing the feasibility and practicality of Al projects. While the potential of Al and machine learning is vast, it is imperative to recognize that not every application seamlessly aligns with the technology’s capabilities or user behavior. In this case, we realized that the level of commitment required from users to effectively train the model about their nuanced preferences was simply too high, resulting in suboptimal outcomes and user frustration.