Spectee Selected for GENIAC, a National Generative AI Foundation Model Development Project by METI and NEDO

Spectee, a resilience-tech startup (HQ: Chiyoda-ku, Tokyo; CEO: Kenjiro Murakami), announced its selection for the 4th phase of GENIAC (Generative AI Accelerator Challenge), a national project conducted by the Ministry of Economy, Trade and Industry (METI) and NEDO (New Energy and Industrial Technology Development Organization). GENIAC is a project aimed at strengthening Japan's generative AI development capabilities, and this marks the first time a company has been selected in the field of disaster and crisis management.

■ Development Goals: An Autonomous Next-Generation Disaster and Crisis Management Platform

In this project, Spectee will work on the following developments:

- A domain-specific model (Disaster Detection LLM) that extracts crisis events such as fires, floods, and accidents from real-time unstructured data (e.g., SNS) and determines the severity of the damage.

- A Crisis Management AI Agent that links vast amounts of information received during disasters with user-specific data, such as disaster prevention plans and BCPs, to provide real-time, prescriptive actions.

This will relieve crisis managers from the burden of information verification and dramatically shorten the lead time for initial disaster response, creating an environment where they can focus on core tasks requiring high-level judgment, such as evacuation decisions and adjustments to supply or transportation routes.

■ Background of Development

In recent years, societal uncertainty has rapidly increased due to the intensification of weather-related disasters caused by climate change, geopolitical risks, and pandemics. Under these circumstances, strengthening disaster management systems during normal times is an urgent task for companies and local governments, and the ability to take swift and accurate initial actions during a disaster is crucial to protecting both lives and the economy.

Meanwhile, in disaster response sites, large amounts of information from various sources—SNS, news, and sensors—are flowing in simultaneously. This flood of data exceeds human interpretation capabilities and hinders rapid decision-making. Specifically,現場 (frontline workers), including government officials and procurement managers in the manufacturing sector, are increasingly demanding concrete action proposals based on the gathered information, rather than just reports.

To meet this social demand, Spectee has decided to develop an autonomous, next-generation disaster and crisis management platform that provides end-to-end support, from extracting and interpreting crisis events to offering real-time action suggestions, utilizing its AI technology and over a decade of accumulated unique disaster-related data.

■ Comment from Nana Kato, Director, CRDO, and Head of Advanced Disaster Prevention Technology Development Office

'I am honored to have been selected for GENIAC. In this day and age, where disasters are becoming more severe and social uncertainty continues to rise, simply collecting and visualizing information is no longer enough to protect lives and livelihoods. While the use of LLMs in disaster and crisis management has great potential, there are many development challenges to overcome—such as suppressing hallucinations and improving the accuracy of action suggestions—especially since this field is linked to human life.

An autonomous system where AI supports disaster response is unprecedented globally, making it an extremely challenging field. However, I am confident that this initiative will create 'time' for crisis management teams and have the value of dramatically changing the disaster response timeline.

We plan to integrate the project's outcomes into our platform after user demonstrations in domestic and international local governments and private companies, and expand it globally as a next-gen disaster and crisis management platform originating from Japan. In particular, ASEAN countries, which are currently experiencing rapid growth while facing disaster threats, are regions where Japanese disaster prevention technology can significantly contribute. We are fully committed to this development to prevent as many disaster-related losses as possible.'