At last year’s Snowflake Summit, we previewed a first of its kind relational knowledge graph running entirely within Snowflake using the Native App Framework with Snowpark Container Services. Today, RelationalAI is generally available on both AWS and Azure and helps Snowflake customers accelerate their journey to build intelligent, AI-powered applications directly within their preferred data cloud. We are also announcing new reasoning capabilities for customers to make faster and better decisions.
RelationalAI offers the only relational knowledge graph that’s fully integrated and native to the Snowflake AI Data Cloud. This enables organizations to apply advanced reasoning to infer, discover, predict and optimize all aspects of their business - unlocking a new class of intelligent applications and super-aligned GenAI question answering workflows.
In collaboration with AT&T, we recently demonstrated state-of-the-art text-to-SQL performance on the Spider 2.0 benchmark by combining relational knowledge graphs with large language models. The same approach enables broader text-to-reasoning capabilities with RelationalAI’s built-in rules-based, graph, predictive and prescriptive reasoners - all operating over a shared semantic foundation. RelationalAI not only captures the concepts and relationships that define a business but also infers new knowledge through reasoning, allowing decisions—whether human- or AI-driven—to leverage shared relational intelligence with 10x less code and complexity.
Since announcing our public preview at Snowflake Summit 2024, we’ve made significant strides to bring the power of RelationalAI to more Snowflake customers around the world. Over the past year, we’ve focused on expanding availability, deepening our Snowflake integration, and delivering a great developer experience. Some of our improvements include:
General Availability: RelationalAI is officially GA on both AWS and Azure, making it easier to get started in customer’s preferred cloud environment. Expanded Global Reach: Expansion of RelationalAI regional availability from 2 at launch to over 12 regions today—bringing our capabilities closer to where customer data lives. Deeper Snowflake integration - Over the past year, RelationalAI has added improvements in performance and end-to-end security through deeper integration with the Snowflake platform. We have also elevated the developer experience by integrating with Snowflake Notebooks and graph visualization support.
Continuing on our journey, at this year’s Snowflake Summit we are proud to introduce new features and capabilities that are now available for customers.
Next-Gen LLM Question Answering with Text-To-Reasoner (Early Access): RelationalAI extends question answering based on retrieval-augmented generation (RAG) and text-to-SQL paradigms with a new text-to-reasoner approach. This will make it possible to leverage RelationalAI’s suite of reasoners in answering questions that are essential for decision making, i.e., what’s going to happen and what to do about it. In a joint submission with AT&T for the Spider 2.0 real-world text-to-SQL benchmark, this technique achieved top of the leaderboard results as of May 30, 2025. Interoperability with Snowflake Semantic Views (Early Access): With interoperability between Snowflake Semantic Views, organizations can now apply business semantics from the relational knowledge graph to increased accuracy for Cortex Analyst and rich dimensional models for BI. This interoperability helps teams drive consistency, accelerate decision making and power intelligent applications with a shared semantic foundation. Integrated Prescriptive Reasoning (Preview): Apps can now use mathematical optimization solvers to compute optimal decisions using clearly defined constraints and objectives. With semantic awareness, applications and AI-assisted workflows can now reason over data for complex domains, such as supply chain planning to balance inventory, cost, demand, and delivery constraints. Expanded Support for Graph Reasoning: Added support for a new egonet graph algorithm (Preview) and new pathfinding capabilities (Early Access), enabling subgraphs extraction and traversal of the relational knowledge graph to find all and shortest paths that meet specified criteria. Integrated Predictive Reasoning with Graph Neural Networks (Early Access): Support for graph neural networks (GNNs) enables applications to learn from both the structure and semantics of data to predict outcomes. This deep learning approach brings new predictive capabilities with higher accuracy and more automation in feature engineering to use cases such as demand forecasting, churn prediction, and product recommendations.
RelationalAI brings business and application semantics directly to enterprise data, enabling reasoning within the Snowflake Cloud. As a native Snowflake app, RelationalAI provides a relational knowledge graph—a semantic foundation that captures business context and executes it through built-in reasoning capabilities, including rules, graph, predictive, and prescriptive reasoners. This allows organizations to build intelligent applications directly in the Data Cloud and unlock the full potential of enterprise AI with relational intelligence.
RelationalAI is at Snowflake Summit at Booth #2412 and featured in multiple sessions throughout the event:
Breakout Session: How AT&T Combines Generative Models with Knowledge Graphs for Better AI Accuracy Breakout Session: Digital Twins Reimagined: A Knowledge Graph AI Approach to Network Analysis | Add to calendar Breakout Session: Observe Inc & RelationalAI: Unlocking AI Power with Snowflake | Add to calendar Breakout Session: AI Knowledge Graph Innovation Drives Life Sciences Media Planning and Strategy | Add to calendar
If you are a Snowflake customer you can get started today by accessing the RelationalAI listing in the Snowflake Marketplace.