BlitzGraph
Supabase for graphs, built for LLM agents
What it does
BlitzGraph is a graph database service designed for LLM agents, positioning itself as "Supabase for graphs." It offers a typed JSON query language called BQL (Blitz Query Language) that agents compose programmatically, avoiding SQL strings. Key features include multi-kind entities (e.g., a User can also be Admin and Moderator simultaneously), bidirectional relationships with O(1) reverse lookup, built-in full-text search (BM25), referential integrity with cardinality constraints and onDelete policies, and business logic (validations, computed fields) defined in the schema. It also provides agent sandboxes (isolated subspaces) and supports MCP (Model Context Protocol) for integration with agents like Claude and Codex.
Who it is for
BlitzGraph is built for developers building AI-native backends, particularly those working with LLM agents. It targets teams that need a graph database with strong relationship support, agent-composable queries, and schema flexibility. It is especially relevant for applications involving agent memory, multi-entity systems, and complex relationship traversals. The product is in public beta and backed by Y Combinator.
Why it matters
Traditional databases (SQL, document stores) force developers to model reality in tables or flat documents, leading to complex joins, migrations, and agent-unfriendly query languages. BlitzGraph aims to solve this by providing a graph-native engine where entities can have multiple kinds, relationships are bidirectional, and queries are structured JSON that agents can generate correctly. This reduces the cognitive load on both developers and AI agents, enabling more natural data modeling and faster development of AI-powered applications. The built-in search, validation, and business logic further reduce the need for external services.
Launch signal
BlitzGraph is currently in public beta and backed by Y Combinator. It was featured on Hacker News as a Show HN post titled "Show HN: BlitzGraph – Supabase for graphs, built for LLM agents." The website offers a live demo with sample data and examples, and invites users to start building for free or book a demo.
Brand and naming
The name "BlitzGraph" combines "blitz" (suggesting speed and action) with "graph" (the core data structure), effectively hinting at fast graph database capabilities. The tagline "Supabase for graphs, built for LLM agents" clearly positions it as a developer tool for AI-native backends, though the name alone doesn't convey the AI/agent focus. The domain blitzgraph.com is short and brandable, enhancing memorability. Overall, the naming is distinctive and relevant, but the positioning relies heavily on the tagline to clarify the AI-agent angle.
Founder
lveillard
Related
Get more like this in our weekly newsletter.