Are LTMs the next LLMs? This new type of AI can do what large-language models can’t

Large-language models (LLMs) have taken the world by storm, but they’re only one type of underlying AI model. An under-the-radar company, Fundamental, is set to bring a new type of enterprise AI model to the masses: large tabular models, or LTMs—which could have an even bigger impact for businesses.

What are LTMs?

A major difference between LLMs and LTMs is the type of data they’re able to synthesize and use. LLMs use unstructured data—think text, social media posts, emails, etc.

LTMs, on the other hand, can extract information or insights from structured data, which could be contained in tables, for instance.

Since many enterprises rely on structured data, often contained in spreadsheets, to run their operations, LTMs could have an immediate use case for many organizations.

What does Fundamental do?

San Francisco-based Fundamental, founded roughly 18 months ago by CEO Jeremy Fraenkel, has made a public LTM model, NEXUS, which will allow organizations to tap into that data to make predictions and forecasts.

The data types in the mix could include customer behavior, information from various sensors, or myriad other things—but again, it’s all locked up in rows and columns.

“If you look at what LLMs have done with unstructured data, it’s been amazing. But it only covers 20% of [overall] data,” Fraenkel says. “That’s the opportunity we’re going after.”

It’s potentially a big deal, because Fraenkel says that roughly 80% of enterprise data used by companies to make predictions and decisions is structured—meaning that it’s on private servers in columns and rows, not really usable by LLMs. 

“You can try things with LLMs, but they’re not really adapted to do it,” Fraenkel says. “They don’t work well with the structured data. They can work with, say, 100,000 rows. But a bank might have tens of billions of rows of data,” which can overwhelm the model. Fundamental’s aim is “the ability to make better predictions” using that structured data. 

Fundamental is also announcing that it’s closed a $225 million Series A funding round. The round was led by Oak HC/FT, and included participation from Battery Ventures, Valor Equity Partners, and Salesforce.

And it’s already worked out some big partnerships, too. That includes one with Amazon Web Services, meaning AWS customers can buy and deploy NEXUS directly through AWS dashboard, and even make payments using Amazon credit, and it’s available today. “We’ll be fully integrated with AWS,” Fraenkel says. AWS customers will have access to Fundamental’s model through their existing contracts, and “any company can use it out of the box.”

Annie Lamont, the founder and managing partner at VC firm Oak HC/FT, which led Fundamental’s Series A round, says that at first, she was “a little skeptical,” but that was soon replaced by excitement as to what the company could be capable of. 

“Weren’t these LLM companies, with endless capital, going to do this? They’re not. They’re different,” she says. “We knew that LLMs are great with unstructured data, but there’s a hole when it comes to structured data—we hadn’t heard of anybody solving the problem.”

“Nobody has commercialized [this type of AI model] for enterprise, so they have a good head start,” she adds.

As for what’s ahead? Deployment, adoption, and proliferation, Fundamental hopes. And if LTMs take off as LLMs did, there’s a very high ceiling: “A few years from now, every Fortune 50 will need to rely on these models to make better business decisions,” Fraenkel predicts.