Startup introduces ‘large tabular model’ for spreadsheet data

by
0 comments
Startup introduces 'large tabular model' for spreadsheet data

Fundamental, a San-Francisco-based startup that developed a way to make predictions from enterprise data, recently emerged from secrecy with a $255 million funding round.

The company has raised a $30 million seed round and a $225 Series A led by Connecticut’s Oak HC/FT, with participation from other venture capital organizations including Valor Equity Partners, Battery Ventures, Salesforce Ventures, and Heitz Ventures, as well as some angel investors. Tangle AI CEO Arvind Srinivas.

At the heart of the fundamental proposal is an AI tool that can understand spreadsheets, an area where larger language models struggle. The company is presenting its innovation in the form of a large tabular model called Nexus.

“We built a generalized base model specifically to leverage the world’s most valuable data: the billions of tables that underpin predictions in every enterprise, across every vertical,” said CEO and co-founder Jeremy Frankel. Press release. “Nexus is the OS for business decisions.”

Connected:Cloud paves a path for NASA’s rover on Mars

According to the company, LLM and related architectures are typically optimized for unstructured, sequential data such as text, images, and video. They are less suitable for tabular data, which is composed of non-sequential, nonlinear relationships, including data commonly found in spreadsheets and tables used in large enterprises.

Nexus aims to solve this problem. Fundamentals said the tool was created by alumni of Google’s DeepMindDeveloped from the ground up and trained on “billions of tabular datasets” Amazon SageMaker HyperPodIt provides the ability to understand non-linear relationships and interactions present in rows and columns.

The company also claims to be easily integrated into the customer data stack. Once connected, the tool automatically starts learning the patterns and structures of the relevant data without any manual training.

“It is difficult to overstate the importance of fundamental models – the benefits of the deep learning revolution have yet to be seen in structured, relational data,” Anne Lamont, managing partner of Investors Oak HC/FT, said in the press release. “The ability to predict anything from financial fraud to hospital readmissions to energy prices enables the company to support nearly every industry and sector.”

Fundamental has also partnered with Amazon Web Services that will allow customers to deploy Nexus in their AWS environments.

Related Articles

Leave a Comment