Last Updated: 4 February 2025
Artificial intelligence has made remarkable strides in processing text and images, yet structured tabular data remains largely overlooked. This type of data—found in databases, spreadsheets, and CSV files—forms the backbone of commerce, powering product catalogs, customer information, and transactional records.
Neuralk-AI, a French deep-tech startup, aims to change this landscape by developing the first Tabular Foundation Model for Commerce.
Unlike large language models (LLMs) that excel at understanding unstructured data, Neuralk-AI's model is specifically designed to process tabular datasets with high precision, enabling businesses to extract actionable insights.
Current AI models, including LLMs, struggle with structured data due to their inherent design. While they perform well in natural language processing and search tasks, they lack the precision required for tabular data representation. As a result, businesses relying on structured datasets face inefficiencies in tasks such as catalog cleaning, fraud detection, and demand forecasting.
Neuralk-AI's approach seeks to fill this gap by creating AI models that can natively understand tabular data. This innovation has the potential to revolutionize commerce, where companies already maintain extensive databases but lack tools to fully leverage them.
The startup's solution offers an API-driven model tailored for data science teams in the retail and e-commerce industries, helping them automate workflows and improve decision-making.
Neuralk-AI recently announced a $4 million funding round led by Fly Ventures with participation from StemAI and notable angel investors, including Thomas Wolf (Hugging Face), Charles Gorintin (Alan), and Philippe Corrot (Mirakl). This funding will accelerate the development of Neuralk-AI’s technology and expand its collaborations with commerce leaders.
The company is currently testing its model with major retailers such as Auchan Retail, E.Leclerc, Mirakl, and Lucky Cart. Over the next few months, Neuralk-AI plans to release its first public benchmark, allowing it to compare performance against state-of-the-art AI models for structured data processing.
“This funding enables us to advance our research and accelerate the development of our AI model tailored for commerce, helping brands and retailers remain competitive in a fast-changing market.”
Antoine Moissenot, Neuralk-AI’s co-founder and CEO
Neuralk-AI’s model unlocks multiple business applications, including:
Unlike traditional machine learning approaches, which require significant manual intervention, Neuralk-AI's foundation model enables businesses to integrate structured data processing seamlessly into their workflows.
While Neuralk-AI is starting with commerce, the potential applications for tabular AI extend beyond retail. Industries such as finance, healthcare, and logistics also rely on structured datasets for decision-making. By focusing on representation learning, Neuralk-AI aims to set a new standard for how businesses leverage structured data across multiple sectors.
“LLM editors aren’t our competitors. While they develop similar models, they target different data types. We’re the first and only ones offering a model specifically designed and trained for tabular data representation, giving us a significant accuracy advantage.”
Antoine Moissenot
With an expanding team of machine learning experts and a clear roadmap, Neuralk-AI is positioning itself as the global leader in tabular data AI. As businesses continue to embrace AI-driven solutions, the demand for structured data intelligence is expected to grow, making Neuralk-AI’s innovation a crucial breakthrough for commerce and beyond.
