R&D at the heart of the reciTAL solution

With 20% of its teams holding PhDs and more than 20 scientific publications in international conferences and journals, reciTAL’s teams contribute to advancing the state of the art.

Generative AI

Text generation is the fundamental building block behind the current chatbot revolution. It is also at the heart of reciTAL’s R&D, with two PhD theses defended on this topic.

Within the reciTAL platform, text generation enables automatic information extraction without training, using commercial or open pre-trained models, as well as the analysis of complex legal documents with little or no training.

Multimodal AI

Analyzing complex documents requires understanding both their meaning (text) and their layout (image). Multimodal AI refers to models capable of leveraging information across different modalities—here, text and image—and using it to deliver more effective analysis.

Within the reciTAL platform, most models are multimodal and leverage text, image, and even word coordinates to provide the highest level of document understanding.

AI ops

The various models available on the reciTAL platform are hosted and operated by reciTAL on GPUs at OVH in France. Depending on their size, different GPUs are used.

In the case of an on-prem deployment, reciTAL can build the servers and deploy them on site, including for offline use if required.

Contactez-nous

Scientific advisory board

reciTAL is one of a small number of companies that is developing the next generation of natural langage technology

Stuart Russell — Professor of Computer Science at UC Berkeley, author of the standard textbook in the field, former vice chair of the World Economic Forum’s Council on AI and Robotics, fellow of the Royal Society.