Airbnb's search feature previously relied on daily batch jobs to update guest embeddings, leading to outdated search results. This staleness could delay reflecting a guest's recent activity for up to two days. To address this, Airbnb introduced two new Chronon capabilities: Push Mode and Near-real-time (NRT) Model Transform. Push Mode allows streaming jobs to notify downstream consumers immediately when new feature values are available. NRT Model Transform enables model inference to run directly within the streaming pipeline. Combined, these features allow guest activity to trigger immediate embedding updates. This process significantly reduces the delay from two days to under a minute. Offline evaluations showed a +1.67% improvement in NDCG, and online tests indicated a one-third percent increase in uncancelled bookings. These general platform capabilities are now available in the open-source Chronon project. This advancement ensures search results are more relevant and responsive to a guest's immediate interests.