Tabular data is a wide date that features structured data, which generally match a specific row and column. It will be a SQL database, a spreadsheet, a .CSV file, etc.
Although there has been great progress in artificial intelligence used for unstructured and sequential data, these large language models are blurred by the project. They are built to control input tokens to generate a coherent output without having to follow a everlasting structure. The best LLM is also expensive in access via the API interface, or expensive in starting your individual infrastructure in the cloud.
And yet many corporations already have a data strategy with a data storage or Data Lake to centralize all vital data and some scientists from data that may use this data to enhance the company’s strategy.
French startup Neuralk-ai It is an artificial intelligence company, he works on AI models that focus on tabular data. This week, the company has announced $ 4 million funding.
“Data on real value for companies is data that was identified a long time ago, structured in the form of a table and used by scientists of data from these companies to create all their machine learning algorithms”, co-founder of Nueralk-ai scientist director Alexandre Pasquiou said Techcrunch.
Neuralk-Ai believes that there is a likelihood to return to the development of AI models again, but with particular emphasis on structured data. Initially, he plans to supply his model as an API interface for scientists working for industrial corporations, because these data – think about product catalogs, customer databases, basketball trends, etc.
“Today LLM is perfect for searching, natural user interaction and answering questions based on unstructured documents. But it has some restrictions when we return to classic machine learning, which is really based on classic tabular data, “said Pasquiou.
Thanks to neuralk-ai, retailers can automate complex data flows with intelligent deduplication and enrichment. However, they’ll use the company’s models to detect fraud, optimize product recommendations and generate sales forecasts that will be used to administer inventory and product valuation.
Fly Ventures led a company round of $ 4 million, and also participated in Steamai. Several business angels also invested in a startup, comparable to Thomas Wolf from Hugging Face, Charles Gorintin from Alan, Philippe Corrot and Nagi Letaifa from Mirakl.
The team is still actively working on their models. He plans to check with a group of leading French retail sellers and industrial startups comparable to E.Leclerc, Auchan, Mirakl and Lucky Cart.
“In three or four months we will publish the first version of our model and a public reference point, where we will be able to order our model compared to the most modern in this space,” said Pasquiou. “And in September, the idea is to be the best model of tabular foundations in everything related to the learning of the national team.”