ABEJA Reports Results Following Completion of NEDO-Promoted 'GENIAC' Project Phase 3
ABEJA has completed Phase 3 of the NEDO 'GENIAC' project, successfully building LLMs and AI agents tailored for mission-critical business operations.
📋 Article Processing Timeline
- 📰 Published: March 28, 2026 at 17:06
- 🔍 Collected: March 28, 2026 at 21:59 (4h 53m after Published)
- 🤖 AI Analyzed: April 15, 2026 at 08:35 (418h 36m after Collected)
ABEJA, Inc. (Headquarters: Minato-ku, Tokyo; CEO: Yosuke Okada; hereinafter "ABEJA"), which aims to "implement a prosperous world" through human-AI collaboration, announces the completion of its participation in the "Research and Development Project for Strengthening the Foundation of Post-5G Information and Communication Systems / Competitive Generative AI Foundation Models (GENIAC)" (hereinafter "Phase 3"), a project promoted by the Ministry of Economy, Trade and Industry and NEDO to strengthen Japan's generative AI development capabilities.
In Phase 3, ABEJA leveraged insights gained from Phases 1 and 2 to build LLMs and AI agents capable of being utilized in mission-critical business operations where errors are not permitted.
As part of this research and development, ABEJA collaborated with SOMPO Holdings, Inc. and Sompo Japan Insurance Inc. to apply the developed LLM to insurance business operations and verify its operational feasibility.
■ Overview
With the management philosophy of "implementing a prosperous world," ABEJA operates a "Digital Platform Business" that develops, implements, and operates the ABEJA Platform, a foundational system for supporting AI integration in mission-critical tasks. Since its founding in 2012, ABEJA has actively engaged in cutting-edge research and development, including deep learning (since 2012), LLMs (since 2018), and quantum computing (since 2019), incorporating these results into the ABEJA Platform.
ABEJA has historically provided the ABEJA Platform for mission-critical tasks at client companies. Due to the nature of these tasks—which require deep expertise, complex handling, and zero tolerance for error—many companies remain cautious about adopting LLMs. Furthermore, as high-level security is essential for most enterprises today, there are inherent limitations to using general-purpose cloud-based LLMs.
Believing that the application of LLMs to mission-critical tasks can transform industrial structures, ABEJA focused its Phase 3 efforts on researching and developing LLMs and peripheral technologies that can be operated safely and securely in such environments.
■ Results
ABEJA's achievements in Phase 3 are as follows:
(1) Compact LLMs for mission-critical use...