Founded by pre-eminent researchers and executives from companies and academic institutions including Airbnb, Pinterest, LinkedIn, Stanford University, and more, the start-up is taking a next-generation approach to predictive AI. Kumo makes graph learning easy to use – so any company can leverage the power of graph-based AI to better their business.
Kumo, a new SaaS AI platform for the modern data stack that allows businesses to make faster, simpler, and smarter predictions, today announced it has emerged from stealth with $18.5 million in Series A funding led by Sequoia Capital, with additional participation from A Capital, SV Angel, Ron Conway, Igor Perisic (Google), Li Fan (CTO, Circle), Tristan Hardy (CEO, dbt Labs), Sridhar Ramaswamy (CEO, Neeva), Greg Greeley (President & COO, Opentrons), Rob Eldridge (Tapas Capital), David Chaiken (Chief Architect, Pinterest), and Cory Scott (CISO, Confluent). Kumo will use the new funding to continue its hiring efforts, scale its leading AI technology, and invest in R&D efforts to expand its platform and services.
Most companies pay hundreds of thousands to millions of dollars every month to store terabytes of data they collect about their products and customers. Yet these businesses are only able to leverage a tiny fraction of this data for predictive tasks. Even companies with the most sophisticated machine learning teams maintain only a handful of AI models because it’s expensive and time-consuming. Kumo’s predictive technology solves for this by bringing powerful graph algorithms into a simple and elegant predictive AI tool for any company to use.
The core graph learning technology that underpins Kumo’s product has been in development for the past five years through Stanford/Dortmund University labs and PyG open-source software, the world’s most widely used graph neural network learning platform. Kumo’s three founders, Vanja Josifovski (former Airbnb CTO Homes, Pinterest CTO, Google), Jure Leskovec (Stanford professor, former Pinterest Chief Scientist), and Hema Raghavan, (former executive at LinkedIn, IBM, Yahoo) saw a tremendous opportunity to take graph learning expertise from an academic setting and operationalize that research for a broad set of use cases in a more user-friendly SaaS product.
“The world of AI is quickly changing and the technological AI landscape in the next few years will be almost unrecognizable. At Kumo, our mission is to be an innovation machine driving that AI frontier,” said Josifovski, Co-Founder and CEO. “Conventional enterprise AI handles each predictive task separately in a silo. However, the underlying data represents business relationships, interactions, customers, and transactions — and these are highly connected entities. Enterprise data forms a natural graph structure which, if leveraged in predictive modeling, can lead to transformational improvements in accuracy, cost, inference speed, and more. Leveraging this inherent connectedness is Kumo’s magic.”
Kumo’s Founding Team has first-hand experience implementing graph-based AI systems at Pinterest and LinkedIn over the past few years. Co-founder, Jure Leskovec, led the development of graph neural network (GNN) embeddings at Pinterest, which resulted in a next-generation recommender system that demonstrated substantial improvements in recommendation performance.1 These embeddings power dozens of use cases across advertising, personalization, toxic behavior detection, growth, fraud, and more.2 At LinkedIn, Kumo co-founder Hema Raghavan also worked extensively with graph algorithms.3 She observed the impact of GNNs to enrich the knowledge graph of member profile attributes and entities. Graph algorithms had a significant impact on LinkedIn’s People You May Know, skill recommendations4 and several recommendation and ranking problems. As CTO at both Pinterest and Airbnb, Vanja defined AI strategies, pioneering graph learning algorithms for real-world applications. Alongside the Founders, the Founding Engineers have experience using GNNs to improve abuse detection, fraud detection, forecasting, among other business growth and trust problems.
In application, Kumo co-founders saw the incredible power of graph learning for AI and business ROI — and also the incredible effort to implement a single, production-quality predictive model. With Kumo, the team aims to solve this problem by making graph learning easy to use – so any business can leverage the power of graph-based AI.
“For years judicious enterprises have invested in data warehouses and data lakes. Yet, extracting valuable insights from that data requires a ton of work by AI and data science teams. Kumo is using new AI techniques to make extracting insight from data as easy as it was to set up a Snowflake data warehouse,” said Konstantine Buhler, partner at Sequoia. “Vanja, Jure and Hema are some of the world’s leading experts in AI and have recruited a team of brilliant engineers. We couldn’t be more excited to partner with the Kumo team to help customers make the most of their data.”
For more such updates and perspectives around Digital Innovation, IoT, Data Infrastructure, AI & Cybersecurity, go to AI-Techpark.com.
