Neuralk-AI is developing AI models specifically designed for structured data, a move that could revolutionize how companies handle and analyze their tabular datasets.
Highlights
What is Tabular Data?
Tabular data, which includes data organized into rows and columns, is commonly found in SQL databases, spreadsheets, and CSV files. Despite its widespread use, this type of data has been somewhat overlooked by current AI models, which are more adept at handling unstructured and sequential data.
The Limitations of Current AI Models
Large language models (LLMs) have made significant strides in processing unstructured data, such as text and images, but they are not optimized for tabular data. These models are designed to manipulate input tokens to generate coherent output without adhering to a fixed structure, which makes them less precise for tasks that require structured data analysis.
Neuralk-AI’s Approach
Neuralk-AI, a French deep-tech startup, is addressing this gap by developing AI models that are specifically tailored for structured data. The company’s focus is on creating a Tabular Foundation Model that can natively understand and process tabular datasets with high precision.
Applications in Commerce
Initially, Neuralk-AI plans to offer its model as an API to data scientists working in commerce companies. This will enable retailers to automate complex data workflows, including tasks such as:
- Catalog cleaning
- Fraud detection
- Optimizing product recommendations
- Generating sales forecasts for inventory management and product pricing
Funding and Collaborations
Neuralk-AI recently secured $4 million in funding led by Fly Ventures, with participation from StemAI and several notable angel investors, including Thomas Wolf from Hugging Face, Charles Gorintin from Alan, and Philippe Corrot from Mirakl. This funding will accelerate the development of their technology and expand their collaborations with major retailers such as Auchan Retail, E.Leclerc, Mirakl, and Lucky Cart.
Future Plans
Over the next few months, Neuralk-AI plans to release the first public benchmark for its model, allowing it to compare its performance against state-of-the-art AI models for structured data processing. By September, the company aims to be the best tabular foundation model in representation learning.
Neuralk-AI is developing AI models specifically designed for structured data, a move that could revolutionize how companies handle and analyze their tabular datasets.