Mitsui Chemicals Develops Patent Chat Platform Using Generative AI

Mitsui Chemicals, Inc. (Tokyo: 4183; President & CEO: HASHIMOTO Osamu) today announced that it has developed a platform equipped with a proprietary generative AI chat system, and has begun demonstration testing of this platform in the company’s various divisions and its R&D Center. The platform is intended to meet the highly specialized needs of the chemical industry, such as reading experiment data tables, understanding chemical structures and more.

Aims are to conclude the demonstration testing within fiscal 2024, then fully implement the platform starting in fiscal 2025.

Mitsui Chemicals aims through this initiative to advance digital transformation (DX) across its business domains, in turn aiding the creation of ideas for new product development and the growth of both product top lines (sales revenue) and market share. In combining generative AI with the chemical expertise of Mitsui Chemicals’ researchers and engineers, the project is expected to bring about new value that could not be achieved with conventional methods.

The newly developed platform will come equipped with three features. First of these is the patent analysis feature, which searches through large amounts of patent data to find information that answers the user’s question. This will be useful in patent searches, as well as in identifying issues and analyzing technology trends.

The second feature is the new use case search feature, which looks through large volumes of patent data and extracts the information required for exploring new use cases. This will aid in exploring new use cases and in suggesting use case candidates.

Third is the sales proposal feature, which looks through both internal data and web information and extracts any information related to sales activities. This can be used for tasks such as conducting market research, suggesting new sales leads and analyzing the competition.

Demonstration testing carried out with the cooperation of Mitsui Chemicals’ divisions and R&D Center has found that using these features in patent searches and in searching for new use cases reduces the time required for these tasks by 80 percent. Particularly substantial optimization (compared to conventional methods) can be expected when using these features to analyze technical documents that include chemical structure formulas and data tables. The sales proposal feature will be capable of obtaining and analyzing not only data in the form of written language but also data tables, the structural formulas of chemical compounds and more, helping meet the complex data processing needs unique to the chemical sector.

Mitsui Chemicals intends to use the information acquired through this initiative to actively pursue industry–academia ties and collaboration between companies, contributing in turn to the development of the chemical industry as a whole.