Mixpeek

Mixpeek turns unstructured media and documents into data that software and AI agents can search and use. In its managed indexing path, connect object storage so files can be processed into features such as video scenes, faces, OCR text, transcripts, and embeddings. Those extracted fields can be filtered, ranked, joined, and queried through retrievers that combine multiple retrieval stages. Teams with existing embeddings can instead use Mixpeek’s vector store, which supports dense, sparse, and BM25 search on object storage. The platform also offers typed feature extractors, integrations for storage and media services, and embeddings from a range of models. Example uses include finding moments in video libraries, searching recordings by transcript or speaker, document layout retrieval, visual product search, and content review. Mixpeek is aimed at developers and technical teams building multimodal search, retrieval, and content-processing systems. The pricing path depends on whether a team supplies vectors or raw files: standalone vector storage begins with a monthly platform price plus usage, while managed indexing charges for processed content and includes a processing allowance with its entry plan.