+Sonae AI

“+Sonae AI” builds on the existing “+Sonae Algorithm,” which provides action prompts based on real-time disaster occurrence and progression. By adding AI analysis of each user’s personal attributes, current time and location, and real-time trends from various open data sources, it delivers action prompts that are even more tailored to the individual. In response to user questions, the system selects and narrows down the most suitable action prompts from a library of 700 content items.

Recommended actions are generated by combining real-time data—such as predicted rainfall and flooding risk, power outages, and other infrastructure information—with the “+Sonae Algorithm” (patented in 2020), which suggests disaster responses in a time-series manner based on user attributes. These attributes include family composition, presence of pets, elderly individuals, pregnant women, and other sensitive information, as well as time and GPS information registered on the device.

When users submit questions via text or voice, a large language model analyzes them. It extracts keywords and matches them against the titles and descriptions of 700 types of action-prompt content, narrowing down the most relevant recommendations.

“+Sonae AI” supports users in deciding what to do not only during natural disasters but also in a wide range of life-risk situations—such as infrastructure failures, car accidents, crimes, and sudden illnesses— by presenting simple, easy-to-understand illustrative and text-based guidance, with 700 types currently available.

Credit

WORKS