Neuralk-AI, a French deeptech startup building a Tabular Foundation Model specialised for Commerce, has raised $4M in funding.

The round was led by Fly Ventures with participation from StemAI, Thomas Wolf, Charles Gorintin, Philippe Corrot, Nagi Letaifa, Vincent Luciani , Saturnin Pugnet, Xavier Perret, Julien Launay and Jean-Louis Quéguiner.

The company provides specialised models to do representation learning on commerce tabular data, such as product catalogs customer information, or shopping cart histories. this allows for enhanced predictive analytics, personalizationi, increased data quality and operational efficiency.

Tabular data lies at the heart of enterprise operations, and is overrepresented in decision-making across industries like finance (e.g. transaction records, account balances), healthcare (e.g. patient information, test results) or retail (e.g. product catalogs, sales data, inventory levels). While it represents only around 20% of data volume, it contains the majority of a company’s value, capturing critical expertise, operational insights, and business know-how.

It does this by providing API access to a representation model that processes tabular data as input and returns the finest numerical representation to compute the given prediction. Neuralk-AI is already collaborating with leading names in the retail industry like E.Leclerc, Auchan, Mirakl, and Lucky Cart.

This funding will enable the company further develop its Tabular Foundation Models. “The technology underpinning tabular data has barely evolved in decades, particularly in retail where structured datasets drive every key decision,” said Alexandre Pasquiou, Co-Founder and CSO of Neuralk-AI. “Our foundation model unlocks hidden potential within this data, empowering businesses to move beyond basic analytics to intelligent systems that deliver real impact.”

“Neuralk-AI’s foundation model is very exciting,” said Thomas Wolf, CSO at Hugging Face. “They unlock the untapped potential of tabular data, working toward establishing new standards for extracting intelligent, actionable insights from structured datasets.”