AI Datastore.
The datastore layer for AI workloads on Origon AI Cloud — one engine for objects, knowledge graph, vector search, full-text retrieval, sessions, lineage, and persistent memory.
S3-compatible object API
Use existing S3 tools and SDKs without changing application code.
Standard APIs
REST and SDK access without a proprietary client.
MCP-native agent access
Agents read and write through MCP.
Full IAM
Fine-grained identity, roles, and access policies.
Artifact storage
Store model outputs, files, and build artifacts.
Session workspaces
Isolated workspaces for each agent session.
Retrieval that stays attached to the data.
Knowledge graph
HNSW vector search
BM25 full-text retrieval
Multi-signal ranking
Persistent memory across sessions
Lineage and audit trail
One datastore across the runtime.
Agents and applications use the same datastore for artifact storage, enterprise knowledge, retrieval, session state, and memory. Knowledge access is logged and traceable. The graph is derived from object and session content, so memory and retrieval stay attached to the underlying data instead of living in a separate sidecar.
