Neuralk-AI, a French startup specializing in artificial intelligence, has embarked on an ambitious project to develop AI models specifically tailored for tabular data. With a strategic focus on structured data—information neatly organized in rows and columns—the company seeks to revolutionize the retail sector. They plan to collaborate with leading French retailers and commerce startups, including E.Leclerc, Auchan, Mirakl, and Lucky Cart.
Supported by $4 million in funding from Fly Ventures and SteamAI, along with investments from notable business figures such as Thomas Wolf from Hugging Face, Charles Gorintin from Alan, and Philippe Corrot and Nagi Letaifa from Mirakl, Neuralk-AI is poised for significant advancement. The company aims to provide its innovative model as an API for data scientists working within commerce companies, ultimately enhancing their analytical capabilities.
Neuralk-AI’s technology focuses on automating complex data workflows, including smart deduplication and enrichment. This approach reflects their belief that revisiting AI model development with a concentrated emphasis on structured data holds immense potential. Unlike Large Language Models (LLMs) that excel in processing unstructured documents, tabular data presents unique challenges and opportunities best addressed through specialized AI models.
“Data with real value for companies is data that was identified a long time ago, structured in the form of a table, and used by the data scientists of these companies to create all their machine learning algorithms,” – Alexandre Pasquiou
Neuralk-AI's objective is to become the leading tabular foundation model in representation learning by September. The initiative promises to set a new benchmark in how structured data is utilized within machine learning algorithms.
“Today, LLMs are great for search, natural user interaction, and answering questions based on unstructured documents. But it has some limitations the moment we go back to classic machine learning, which is really based on classic tabular data,” – Alexandre Pasquiou
The company plans to release the first version of its model within three to four months. This release will include a public benchmark enabling the comparison of their model against current state-of-the-art solutions.
“Within three or four months, we’ll release the first version of our model and the public benchmark on which we’ll be able to rank our model compared to the state-of-the-art in this space,” – Alexandre Pasquiou
Leave a Reply