Neuralk AI

Neuralk AI is a French deep-tech startup addressing the often-overlooked area of structured data within the artificial intelligence landscape. Founded in 2023 by CEO Antoine Moissenot, a graduate of École Polytechnique, and CSO Alexandre Pasquiou, who holds a PhD in neuroscience, the company is developing foundation models specifically for tabular data. The core idea for the technology originated from Pasquiou's doctoral work and his exploration of its practical business applications. The company asserts that while large language models (LLMs) excel with unstructured text and images, they lack the necessary precision for the structured, tabular data that forms the backbone of most enterprise operations.

Neuralk AI's primary offering is a Tabular Foundation Model specialized for the commerce sector, delivered to clients via an API. This model processes structured datasets like product catalogs, customer information, and transaction histories to enable more accurate predictive analytics and automate data processing. For its enterprise clients, which include data science teams in the retail and e-commerce industries, the platform aims to improve operational efficiency. Key applications include automated data cleaning and deduplication, enhanced product recommendations, fraud detection, and more precise sales forecasting for better inventory and pricing strategies. The business model appears to be subscription-based, determined by the client's use case and data volume.

In February 2025, the Paris-based startup, which is housed at Station F, announced a significant milestone, having secured $4 million in seed funding. The round was led by Fly Ventures and included participation from StemAI and prominent angel investors from the tech community, such as Thomas Wolf, co-founder of Hugging Face. This capital is intended to accelerate research, advance the development of its AI model for commerce, and expand its team. Neuralk AI is already collaborating with major European retailers and e-commerce companies like E. Leclerc, Auchan, and Mirakl to test and refine its technology.

Keywords: tabular data, foundation model, structured data AI, predictive analytics, commerce AI, data representation, Graph Neural Networks, data cleaning, retail technology, e-commerce optimization, sales forecasting, fraud detection, personalization engine, AI embedding, machine learning, data science API, deep-tech, Antoine Moissenot, Alexandre Pasquiou, Station F