Back to products
Beta
Citadea Memory
A hardware-accelerated vector database for agents
Long-term memory for agents that does not leave the EU — vector search in under a millisecond.
Overview
A distributed, hardware-accelerated vector database natively integrated into the storage layer. Sub-millisecond vector search, multi-agent persistent memory, native RAG integration and snapshots — long-term memory for enterprise AI agents, hosted entirely in the EU.
A vector store wired into the platform
Memory is a distributed, hardware-accelerated vector database integrated into the storage layer. Sub-millisecond search, multi-agent persistent memory, native RAG integration and snapshots — all hosted 100% in the EU, reachable over REST and gRPC.
Compared to Pinecone and Weaviate.
What's included
- Sub-ms vector search
- Multi-agent persistent memory
- Native RAG integration
- 100% data in the EU
- Snapshots and restore
- REST and gRPC API
Quickstart
$ citadea memory query <coll> --text "..."Benchmarks
< 1 ms
Vector search
100% EU
Residency
Use cases
Long-term agent memoryEnterprise RAGAI knowledge baseMulti-agent context
Integrations
LangChainLlamaIndexRESTgRPC
Frequently asked
100% in the EU, integrated into the sovereign storage layer with snapshots and restore.