The field of vector search and indexing is moving towards more semantic and context-aware approaches. Researchers are exploring ways to improve the diversity and richness of search results, beyond traditional nearest neighbor searches. This includes the development of new retrieval paradigms, such as semantic compression, and the integration of graph structures to enable multi-hop, context-aware search. Additionally, there is a focus on improving the efficiency and scalability of indexing systems, including the development of dynamic and adaptive indexing methods. Notable papers in this area include: Beyond Nearest Neighbors, which introduces a new retrieval paradigm for semantic compression and graph-augmented retrieval, and CleANN, which proposes an efficient and dynamic graph-based indexing system. AutoIndexer is also noteworthy, as it presents a reinforcement learning-enhanced index advisor for scaling workloads.