orneryd/NornicDB

Nornicdb is a distributed low-latency, Graph+Vector, Temporal MVCC with all sub-ms HNSW search, graph traversal, and writes. Using Neo4j Bolt/Cypher and qdrant's gRPC means you can switch with no changes while adding intelligent features like schemas, managed embeddings, reranking+llm, GPU accel, Auto-TLP, Policy-based Memory Decay, and MCP server.

What it solves

NornicDB is a unified database designed for AI-native workloads that require graph traversal, vector retrieval, and historical truth tracking in a single system. It eliminates the need to maintain separate databases for graph data (like Neo4j) and vector embeddings (like Qdrant), reducing architectural complexity and improving retrieval performance for Graph-RAG and agent memory systems.

How it works

NornicDB integrates graph, vector, and ledger semantics into one execution engine. It is Neo4j-compatible, supporting the Bolt protocol and Cypher query language, while also providing Qdrant-compatible gRPC interfaces. The system uses Snapshot Isolation via MVCC (Multi-Version Concurrency Control) to provide historical "as-of" reads and repeatable reads within transactions. To optimize performance, it employs hardware-accelerated execution paths (Metal, CUDA, Vulkan) and specialized streaming executors for common query shapes.

Who it’s for

It is built for developers creating knowledge systems, AI agent memory architectures, and Graph-RAG pipelines who need a single source of truth that supports both semantic search and complex relationship mapping.

Highlights

  • Unified Engine: Combines graph, vector, and temporal ledger capabilities without requiring sidecar databases.
  • Broad Compatibility: Supports Neo4j's Bolt/Cypher and Qdrant's gRPC protocols.
  • Hardware Acceleration: Native pathways for Metal, CUDA, and Vulkan to speed up semantic and graph workloads.
  • Knowledge-Layer Scoring: Implements profile-driven decay and promotion policies for managing episodic and durable memory.
  • Temporal Truth: Supports tritemporal facts and historical reads via MVCC.
  • Auto-Relationships: Automatically creates connections based on embedding similarity, co-access patterns, and temporal proximity.

Related

  • Dispatch
    LatticeDB: An Embedded Single-File Graph Database with Vector and Full-Text Search

    LatticeDB is an embedded, single-file property-graph database that integrates native vector similarity search and BM25 full-text indexing into a single query layer for local-first AI and RAG applications.

  • Project
    rush-db/rushdb

    A memory layer for AI agents that combines key-value, vector, and graph databases into a single schema-free API with automatic embedding and relationship inference.

  • Project
    nuclia/nucliadb

    NucliaDB is a hybrid AI search database that combines vector, full-text, and graph indexes to store and search unstructured data.

  • Project
    ArcadeData/arcadedb

    A high-performance multi-model DBMS that supports graph, document, key-value, time-series, and vector data in a single ACID-compliant engine.

  • Project
    HelixDB/helix-db

    A graph-vector database built in Rust that consolidates relational, vector, and graph data into a single platform for AI memory and knowledge graphs.