Numbers Station, a startup that’s using large language models (LLMs) to power its data analytics platform, is launching its first cloud-based product today: the aptly named Numbers Station Cloud, which is now in early access. With this service, virtually any user in an enterprise can analyze their internal data using Numbers Station’s chat interface.
Several similar tools give attention to translating natural language queries right into a database language like SQL. Nevertheless, the Numbers Station team argues that this approach has limitations, partially since the generic LLM doesn’t have an understanding of how a given company operates, the way it structures its data and the way people contained in the company seek advice from specific data objects.
Image Credits: Numbers Station
As Numbers Station co-founder and CEO Chris Aberger told me, he’s somewhat bored with talking about how the service allows users to “chat with their data,” because there may be a lot noise around that. “But the upper level thing of business executives, non-technical users, having questions they wish to ask, after which getting answers over these classic structured data sources, is de facto where things have led,” he told me. “There’s loads of data modeling, data plumbing around these foundation models and huge language models to make them work.”
For Numbers Station, this has meant spending loads of engineering resources on constructing what the corporate calls its semantic catalog. That catalog is actually an robotically curated source of an organization’s metrics and definitions. That catalog is particular to each company (and never shared between them). Aberger described the catalog as “a beastly thing” that, for instance, ensures that the model’s definition of “recurring revenue” is aligned with the corporate’s usage of that term.
While Numbers Station’s platform sits on top of a set of very specialized LLMs and machine learning models, it’s this catalog that holds all the things together. As Numbers Station co-founder and chief scientist Ines Chami told me, the team had originally underestimated the challenge of constructing out that a part of the platform.
“It goes back to classical [machine learning] and classical data engineering: How do I create a representation of information that the model can actually use to reply those questions,” she told me. “Because there’s no way a model goes to grasp all those metrics, all those things that business users ask about.” Even humans don’t understand every query immediately, in spite of everything, and the model has to show those vague questions into very concrete queries. Numbers Stations’ research shows that its approach leads to significantly improved precision in comparison with more traditional text-to-SQL pipelines.
Image Credits: Numbers Station
While the corporate is launching this chat service today, the general vision is significantly larger.
“What we’re doing, fundamentally, is constructing an AI platform for analytics,” Aberger said. “That is one in every of the applications […]. There’s an even bigger, broader play that we’re still working on as an organization, which goes after a bunch of various data problems that sit on top of here, examples of that are: How do I enrich my data with third-party data sources? How do I do a few of these more classic algorithms like fuzzy matching, etc.? There’s almost an infinite variety of spokes which you could construct off of on this platform.”
The corporate has already signed up several Fortune 500 customers, including the likes of worldwide real estate services firm Jones Lang LaSalle. “Numbers Station is on the innovative of enterprise AI for structured data,” said Sharad Rastogi, the CEO of Work Dynamics Technology from Jones Lang LaSalle. “We’re impressed by Numbers Station’s trusted and fascinating platform. It constantly learns as we use it, enabling our data teams discover and confirm hypotheses for driving impactful business outcomes.”
