Hey HN, we want to share HelixDB helix db, a native Graph-Vector Database project we are building to solve the fragmentation in modern AI infrastructure. By combining semantic similarity with complex relationship querying, we provide a unified system for developers building advanced AI agents.

Why a hybrid?

Vector databases are powerful for similarity, but many AI-driven applications need both similarity and relationship queries to succeed. Why not use multiple systems? Maintaining separate databases like Postgres and Pinecone introduces massive latency and synchronization nightmares. We built HelixDB to bridge this gap, allowing developers to handle multi-hop reasoning and semantic search within a single, unified system.

Use Cases

• Supply Chain Optimization: You can vectorize supplier data while mapping complex contractual relationships to identify ripple effects when a single vendor experiences a delay.
• Enterprise RAG: Index vast corporate wikis where you need to perform multi-hop reasoning to connect policies to specific department workflows.
• Molecular Research: Model biological interactions as graph nodes while using vector embeddings to identify structural similarities between compounds.

Technical Deep Dive: Why Another Database?

Many people have a thing against “yet another database architecture,” but we went ahead and did it anyway because we think it makes working with our database so much easier than managing multiple disconnected services. Our LSM-based storage engine, implemented in Rust, provides memory safety and raw speed. Our benchmarking shows that we are on par with dedicated vector stores like Qdrant for similarity search, and our graph traversal is up to three orders of magnitude faster than Neo4j when co-located with vector indices.

Next Steps

We are excited to see what the community builds. Here is a video runthrough of the architecture: helix db. If you’d like to try it out in a simple RAG demo, you can follow our guide here: docs.helix db.com. Many thanks! Comments and feedback welcome!