Neo4j brings graph intelligence to Snowflake’s data platform
Neo4j and Snowflake are offering integrated tools that allow companies to build knowledge graphs directly on Snowflake data and run graph analytics. The partnership aims to give artificial intelligence enterprise context without the time-consuming process of moving data.
- Virtual Graph builds a knowledge graph from Snowflake data without moving it.
- The system converts Cypher queries into SQL statements.
- Graph Analytics for Snowflake offers more than 65 graph algorithms.
- Native Neo4j databases target real-time, low-latency tasks.
- Audience Acuity reduced processing time from several weeks to less than 24 hours.
Building a knowledge graph directly on Snowflake data
According to a November 3 announcement from Neo4j, the partnership includes Virtual Graph for Snowflake, Graph Analytics for Snowflake and Aura Data Importer for Snowflake. The tools target different use cases within the Snowflake AI Data Cloud environment.
Virtual Graph for Snowflake is currently available in public preview, with general availability planned for November 16. The tool creates a knowledge graph based on an organization’s data stored in Snowflake, without moving the data. Users can run Cypher queries against the data warehouse, while the system automatically converts them into deterministic SQL statements and returns the query results as a graph.
The solution preserves existing data governance rules and Snowflake’s computing environment. Neo4j says AI-assisted schema design can produce an initial graph model in a matter of minutes. For users unfamiliar with Cypher, Copilot can also convert natural-language questions into graph queries.
Analyzing data with SQL without moving it
Graph Analytics for Snowflake is already available on the Snowflake Marketplace. The serverless, fully managed application uses SQL to create in-memory graphs from Snowflake tables, run algorithms and write the results back to Snowflake tables.
Neo4j offers more than 65 graph algorithms for uncovering relationships and patterns in data. According to the company, the service charges only for the resources used, with no monthly minimum or inactivity fees. Potential applications include supply chain optimization, collusive fraud detection, marketing entity resolution, as well as customer segmentation and next-best-action recommendations.
Snowflake Cortex AI and Neo4j’s graph context can work together to provide natural-language answers to business questions that take relationships into account.
Real-time applications may still require a separate graph database
According to Neo4j, tasks requiring millisecond latency, CRUD operations, ACID transactions or queries across multiple data sources can still use a native Neo4j graph database. Examples include cyberthreat analysis, next-best-action recommendations, root-cause analysis for network and IT failures, and live fraud detection.
Users can migrate relevant subgraphs, known as hotspots, to a Neo4j database. Aura Import Service supports Snowflake as a data source, while AuraDB is Neo4j’s fully managed graph database service.
Neo4j also cites a customer example: Audience Acuity accelerated identity resolution using more than 2 billion connected data signals, reducing processing time from several weeks to less than 24 hours. The company says Virtual Graph for Snowflake could give more than 12 thousand organizations the ability to query Snowflake data as a graph in place.
