Pinecone AI Search
Pinecone vector DB serverless RAG-optimized.
Définition
Pinecone Serverless architecture découplée compute / storage, pay-per-query, auto-scaling. Distance metrics : cosine, dot-product, euclidean. Métadonnées filtering. Hybrid search (sparse-dense). Embeddings managed (Pinecone Inference). SDK Python, Node.js, Java, Go. Free tier 2GB storage. Plus de 5000 entreprises clientes incluant Shopify, Notion, Gong, You.com.
Origine
Pinecone Systems Inc., fondée 2019 à New York, Series A 28M$ 2021, Series B 100M$ 2023 valorisation 750M$.
Usage
Vector DB SaaS dominant. Concurrent Weaviate, Qdrant, ChromaDB (OSS), Milvus (Zilliz Cloud), Vespa.
Termes liés
- Vector database — catégorie.
- RAG — cas usage central.
- Embeddings — inputs natifs.