Kumo unveils world's first Relational Foundation Model (RFM) for instant predictions on enterprise data

Yahoo

20-05-2025

Today, it takes months to build predictive models using machine learning — KumoRFM delivers results in seconds, with no manual effort.
MOUNTAIN VIEW, Calif., May 20, 2025 /PRNewswire/ -- Kumo, a leader in predictive AI, announced today the launch of KumoRFM, which allows businesses to instantly generate accurate predictions, such as item recommendations, identifying customer churn, or detecting fraudulent transactions, directly from their enterprise data. The first-ever foundation model built for relational data, KumoRFM gives organizations of any size an off-the-shelf, cost-effective AI model that eliminates the need to manually build and train separate models for each predictive task. KumoRFM results in 20x faster time to value and delivers 30-50% higher accuracy compared to traditional approaches.
AI has completely transformed how businesses leverage text-based and unstructured data like documents, audio, video, and images. Yet, structured enterprise data — like customer records, transaction histories, and product catalogs — is the backbone of business decision-making, but remains largely untouched by this wave of AI innovation.
"To make predictions and business decisions, even the largest and most cutting-edge companies are using 20-year-old machine learning techniques on the enterprise data inside their data warehouses," said Jure Leskovec, Co-Founder and Chief Scientist at Kumo. "Extending Transformer architecture beyond natural language took significant innovation and investment. We're proud to bring to enterprise data what GPTs brought to text, and at a fraction of the cost."
KumoRFM is the first model to deliver instant predictions on structured enterprise data, without the need for task-specific model training. While language models such as ChatGPT use their understanding of semantic meaning within text to generate the next word, KumoRFM uses its understanding of traits, behaviors, and relationships represented inside data warehouses to predict what business outcomes will be at a different time or in a different scenario. These predictions power critical business decisions, from flagging suspicious transactions to recommending products, personalizing marketing offers, and selecting which ads or content to display. All of which have a direct impact on revenue.
Data science teams, developers, and engineers can connect KumoRFM to their data warehouse through an API, and KumoRFM can begin immediately identifying patterns within the enterprise's data and make business-critical predictions, like customer churn, personalized customer recommendations, and fraud detection. Unlike traditional approaches that require building and training separate models for each such predictive task, KumoRFM is zero-shot, enabling accurate predictions without the need to even train or specialize the model. This shift allows AI and engineering teams to get predictions for different use cases in real-time, letting them explore and ship applications faster. KumoRFM is trained solely on synthetic enterprise-like data, which makes the model compact and inference cost-effective.
"AI tools like chatbots and content generators have shown what's possible with language, but there's a missing piece when it comes to enterprise data, and KumoRFM fills that gap," said Vanja Josifovski, Co-Founder and CEO at Kumo. "The game changes completely when AI connects with business data. That's when we see the needle move. Real numbers, real ROI, and real business impact."
KumoRFM is a pre-trained Relational Graph Transformer model, enabling it to learn across multiple tables of structured enterprise data. KumoRFM works out of the box, delivering instant predictions with accuracy on par with manual models. When fine-tuned to a specific task, KumoRFM delivers 30-50% higher accuracy compared to traditional approaches. The model is built on years of research in Graph Neural Networks (GNNs) and Graph Transformers, a field pioneered by Jure Leskovec and several members of the Kumo team. This core technology, available through an API, already powers Kumo's existing platform, which is used by more than 20 enterprise clients, including DoorDash, Databricks, Snowflake, and Reddit. To get started, visit Kumo.AI.
Follow Kumo
LinkedIn: https://www.linkedin.com/company/kumo-ai/X (formerly Twitter): https://twitter.com/Kumo_ai_teamKumo Research: https://kumo.ai/research/
About Kumo
Kumo transforms how businesses create predictions from their data warehouse, using AI to quickly build high-performing machine learning models that help data scientists better predict user and customer behaviors with best-in-class accuracy. The company was founded by three PhDs who have held executive leadership and academic positions at Pinterest, Airbnb, LinkedIn, and Stanford. Kumo has raised $37 million in funding and is backed by Sequoia Capital. Kumo is reshaping the future of applications, making predictive AI accessible and practical for companies of all sizes. To learn more, visit kumo.ai.
Contact:
Ali GoldbergConcrete Media for [e-mail]
View original content to download multimedia:https://www.prnewswire.com/news-releases/kumo-unveils-worlds-first-relational-foundation-model-rfm-for-instant-predictions-on-enterprise-data-302460899.html
SOURCE Kumo.AI