May 2026 Summaries

Sunny Health developed an AI healthcare concierge using Qdrant to facilitate patient interaction with insurance benefits, provider matching, and appointment booking, aiming to alleviate the complexity of navigating healthcare plans. Initially relying on Postgres, the company faced challenges in managing a massive dataset of providers and insurance plans, which led them to adopt Qdrant for its hybrid search capabilities. Qdrant's Rust-based architecture allows for a combination of hard filters and semantic similarity in queries, essential for the precise demands of healthcare matching, including strict licensing and network adequacy requirements. The transition resulted in a significant reduction in retrieval latency and improved accuracy in provider matching, with Qdrant enabling queries to efficiently resolve detailed patient requests in one step. Sunny Health plans to expand its use of Qdrant for tasks like insurance carrier matching, semantic specialty mapping, and maintaining conversational memory, further enhancing the concierge's ability to streamline healthcare access.

GoPerfect has revolutionized the recruiting process by leveraging Qdrant Cloud to create an agentic workforce that effectively combines AI and advanced search techniques to enhance candidate selection from a vast pool of profiles. The company faced challenges with traditional vector search systems, which struggled to accurately match candidates due to semantic similarities that did not translate into role suitability. By integrating large language models with semantic search, vector retrieval, and structured filtering, GoPerfect developed a system that narrows down 200 million profiles into high-confidence shortlists, significantly improving the acceptance rate from a 30 percent industry standard to 99.993% in internal benchmarks. The hybrid search and multivector representation allow for nuanced candidate evaluation, while the LLM orchestration layer ensures comprehensive and rapid query processing, enabling complex searches to be completed in seconds. This innovation has transformed GoPerfect's recruiting platform from a basic search tool into an agentic experience that offers recruiters a streamlined and accurate candidate selection process, promising further automation and integration advancements in the future.

Qdrant 1.18 introduces TurboQuant, a new vector quantization method developed by Google Research, which enhances compression for production embeddings while maintaining high recall rates. TurboQuant offers a rotation-based quantization approach with different operating points, such as 4-bit, 2-bit, and 1-bit, allowing it to outperform existing methods like Scalar Quantization (SQ) and Binary Quantization (BQ) in storage efficiency and recall. Benchmarks demonstrate that TurboQuant 4-bit provides competitive recall with SQ but at half the memory usage, while TurboQuant 2-bit and 1-bit achieve significantly higher recall than BQ at the same storage levels. The implementation includes enhancements like length renormalization and per-coordinate calibration, ensuring robust performance across various datasets. The migration to TurboQuant requires a simple configuration change and re-indexing, offering a seamless upgrade path for users of SQ and BQ.

Sapu, a biopharmaceutical company focused on developing nanomedicine treatments for hard-to-treat cancers, has leveraged Qdrant's vector search technology to efficiently index 28 million PubMed abstracts, thus significantly enhancing its research capabilities. Initially facing limitations with early vector search solutions, Sapu transitioned to Qdrant Cloud Premium to address stability issues and gain compliance certifications like SOC 2, which are essential for licensing its AI platform to other biotech firms. This move has allowed Sapu to conduct comprehensive searches across biomedical literature, leading to the publication of seven peer-reviewed research papers and facilitating broader internal and external collaborations, including a partnership with robotics company Techforce to integrate AI into lab workflows. The AI platform, which began as a command-line prototype, now supports various research functions, from document retrieval to SOP-aware chatbots, and is set to expand into robotics and edge deployments, underscoring the importance of rapid information retrieval in accelerating therapeutic advancements.

Qdrant 1.18 introduces several enhancements, including TurboQuant, a new quantization method developed with Google Research, which doubles the compression ratio of scalar quantization while maintaining similar recall and speed, and improved memory monitoring via a Web UI and API endpoint to aid in capacity planning. The update allows adding or removing named vectors in a collection's schema without needing to recreate it, simplifying embedding model migration. Audit logging sees upgrades with a new API endpoint for querying logs and support for request tracing IDs, enhancing security reviews and compliance audits. Per-collection API metrics are now available, providing specific insights into response times and error rates, and new strict mode guardrails help prevent system overload by rejecting excessive memory operations and capping batch search requests.