RelationalAI Adds Agentic Decision Intelligence Capabilities to Snowflake AI Data Cloud
SAN FRANCISCO – RelationalAI, a leader in business AI, announced at Snowflake Summit 26 a series of new capabilities for Rel, its decision intelligence system that runs natively on the Snowflake AI Data Cloud.
With these new capabilities, corporate customers can provide decision-making agents with the context, reasoning, and post-training needed to take action on all operations that drive the bottom line, including pricing, supply chain, network operations, and resource allocation.
Generative AI has opened up an incredible amount of software development, but many businesses still see a gap between what AI can do and what they get from the rest of the business. The release introduces a new Rel App, next to instructions and predictions
thinkers, conversational decision intelligence within Snowflake CoWork, and RelationalAI “push-button” post-training. Together, they provide decision-makers with what they need to act with confidence: a contextual business model, thinkers as tools, and post-training to turn a general-purpose model into a business expert.
“Just like humans, agents have difficulty making good decisions,” said Molham Aref, founder and CEO of RelationalAI. “With these capabilities running natively on the Snowflake AI Data Cloud, customers can close the AI value gap by giving their agents the context, tools, and post-training they need to take better action in the face of uncertainty, at machine speed and in business-wide economics.”
The new Rel App captures a shared, governed representation of how a business works: the concepts, relationships, and rules that define how decisions are made. Domain experts can examine the model, trace interactions, ask questions in natural language, and make inferences
decisions, with all interactions based on their data within Snowflake.
The release also includes RelationalAI’s growing library of code agent capabilities, which work across Snowflake CoCo, Claude Code, OpenAI Codex, and GitHub Copilot. Relational customers are already using these capabilities to extend RelationalAI models directly into their preferred development
areas, deepen the context their decision agents use.
The regular availability of a dedicated thinker provides Snowflake customers with a purpose-built tool to solve delayed optimization problems. Predictive Expert applies graph neural networks to enterprise data within Snowflake to predict outcomes such as demand, churn, and
equipment failure. Paired with a deterministic, predictive thinker it gives decision agents a full path from prediction to recommended action in a single workflow, all without moving data to the platform.
The RelationalAI suite of rules, graphs, predictions, and descriptive rules supports complex multi-domain reasoning in a single workflow. Decision agents running on Snowflake can now combine LLM-based reasoning with domain-specific reasoning, with measurable gains in accuracy and significant cost reductions.
As a launch partner in the Open Semantic Interchange (OSI) system, RelationalAI also enables enterprises with existing ontology deployments, such as Palantir, to embed semantic models in Snowflake through OSI and use advanced logic in RelationalAI without any refactoring required.
“At Snowflake, we’re focused on enabling secure, high-performance AI where data resides,” said Amy Kodl, SVP, Worldwide Alliances and Channels at Snowflake. “RelationalAI’s Rel App extends these capabilities by introducing powerful reasoning and semantic modeling.
within the Snowflake AI Data Cloud, which helps customers accelerate the development of intelligent agents and decision intelligence systems.”
RelationalAI also now enables conversational decision intelligence within Snowflake CoWork that allows business users to ask ad-hoc questions in natural language and receive controlled, nature-based answers from RelationalAI’ thinkers directly from private data in the Snowflake AI Data Cloud.
RelationalAI is also announcing a private preview of RelationalAI “push-button” post-training, the ability to customize open source LLMs against specific business data and semantic fields within Snowflake. Specialized business models after training, when combined with earlier models, can solve complex problems at a fraction of the cost, while learning the systems, terminology, and decision logic specific to the business they provide.
