| What it does | Open-source database for AI agents with self-describing schema
| The AI-native Postgres backend for apps, with RAG and agents built in
|
|---|
| Stage | Not given
| Not given
|
|---|
| Founded | Not given
| Not given
|
|---|
| Based in | AM
| Not given
|
|---|
| Pricing model | Freemium
| Freemium
|
|---|
| Pricing | Free plan: 1 GB storage, 3 projects, 1 seat, 200 Smart Searches/month, 1M embedding tokens/month. Pro: $49/month, 20 GB storage then $0.50/GB-month, 10 projects then $3/month each, 3 seats then $15/month each, 2,000 Smart Searches/month then $5 per 1K, 25M embedding tokens/month then $0.20 per 1M. T
| Free: $0/month with pay-as-you-go usage, $10 in free credits on sign-up. Self Serve: $25/month with $25 in monthly credits. Scale: $300/month with $300 in monthly credits. Enterprise: Custom pricing. Unused credits roll over month to month. Compute tiers (Sandbox to Foundry) range from $0.0272/hr to
|
|---|
| Key features | - Schema catalog automatically updated on every write with exact field names, types, values and relationship shapes
- Nested JSON documents converted to linked records in a single commit
- Multiple query languages: SQL, openCypher, Gremlin, ISO GQL, Mongo-style and native SearchQuery on same data
- Hybrid search combining vector (exact and HNSW) and BM25 full-text indexes in one query
- PostgreSQL and Neo4j protocol compatibility via wire protocols
- MVCC snapshot isolation for concurrent reads and writes
- Atomic transactions with write-ahead log durability
- Single Rust binary deployment with no additional services required
- RushDB Cloud managed service with embeddings and MCP hosting
- TypeScript and Python SDKs plus gRPC and HTTP APIs
| - Per-project Postgres with pgvector
- Authentication with RLS and multiple OAuth providers
- Built-in RAG pipeline with 91% OCR accuracy
- Multi-strategy document indexing (vector, BM25, hybrid, tree search)
- Agent runtime with ReAct orchestrations
- Eight built-in tools plus custom HTTP and MCP integrations
- Visual workflow builder with natural-language copilot
- Agent session memory in Postgres
- Real-time capabilities
- Object storage with configurable tiers
|
|---|
| What makes it different | - Agents see exact schema before querying, eliminating guesswork about field names and types
- Smaller AI models achieve higher accuracy when querying with schema awareness
- Filtered schema queries return accurate facet counts and field values matching query results
- One database handles records, relationships, vectors and full-text indexes without sync overhead
- Free open-source engine and managed cloud option with same underlying technology
| - Complete isolation: every project gets its own database, storage, and compute with no shared infrastructure
- RAG pipeline built-in with no separate vector database needed
- Open source self-hosting option with Docker Compose
- SOC 2 and ISO 27001 compliance by default through project isolation
|
| Community votes | 0 | 0 |
| DR (Domain Rating by Ahrefs) | 10 ● no change since last check | 35 ● no change since last check |
| Trust Flow (Majestic) | 4 | 9 |
| Citation Flow (Majestic) | 23 | 32 |
| Referring domains (Majestic) | 66 | 163 |