“In the AI era, we must accelerate the modernization of enterprise systems and safeguard society’s foundations.”
So remarked Junichi Miyakawa, President & CEO of SoftBank Corp. (TOKYO: 9434), during his keynote presentation at SoftBank World 2026, a major event for enterprise customers in Tokyo, held on July 14. In his presentation, he discussed how companies must modernize their systems to prepare for the AI era by shifting from conventional human-centric models to an AI-first approach.
The importance of rebuilding from the ground up
Miyakawa opened his address by looking back at the Great Chicago Fire of 1871. “While the fire devastated much of the city, rather than simply restoring what had been lost, Chicago chose to rebuild from the ground up,” he said. “That decision laid the foundation for the rise of skyscrapers and the development of the modern city. The story offers an important lesson for today's enterprise systems. If we just respond to accidents, disasters or cyberattacks as they occur, we’ll never achieve true resilience.”
“AI’s rapid evolution is reshaping the business environment at an unprecedented speed,” he added. “That’s why enterprise systems can no longer be maintained as extensions of current systems. They must be transformed into resilient, always-on foundations.”
AI-driven cyberattacks no longer a hypothetical scenario
Miyakawa then explained the current cybersecurity landscape. “A cyber model is an AI model that delivers advanced cyber capabilities by relaxing certain restrictions under controlled safety measures, “ he said. “It can connect the dots across vulnerabilities in enterprise systems—such as outdated software, configuration errors and weak authentication—and even analyze the likely paths an attacker could take. In other words, a cyber model enables us to see not only where vulnerabilities lie, but also how an attacker could gain entry and what they are likely to target.”
Miyakawa said that SoftBank used OpenAI's cybersecurity technologies to assess its internal systems, and identified 10,500 vulnerabilities across 700 systems for which it maintains its own source code.
Leveraging this know-how, SoftBank and SB OAI Japan rolled out Patching as a Service—an OpenAI-powered cybersecurity solution that provides end-to-end support, from vulnerability assessments to patch deployment. In the initial rollout, Miyakawa noted that assessment requests were received from 137 companies, and among those whose assessments have already been completed, approximately 280 vulnerabilities were identified per 10 million lines of source code. The findings demonstrated that the risk posed by cyberattacks extends far beyond companies' core business systems.
“The threat of AI-driven cyberattacks is no longer a hypothetical scenario,” Miyakawa said. “Once cyber models become open—meaning their code and other components are made publicly available, allowing third parties to use and modify them—launching cyberattacks will become dramatically easier. Even individuals without specialized expertise will be able to carry out highly sophisticated attacks.”
Modernization for the AI era will contribute to cyber resilience
“I believe what companies need going forward is a comprehensive review of their entire enterprise systems, and a replacement of source code so it’s compatible with the latest generation of AI,” Miyakawa said.
Miyakawa described this process as “modernization for the AI era.”
“Modernization for the AI era means replacing the traditional, human-centric assumptions underlying systems development with an AI-first approach. In other words, it means transforming system design, operations, and source code so they can withstand AI-driven cyberattacks.”
“Traditionally, enterprise systems have been built around people,” he added. “When multiple teams and multiple vendors are involved in system design, issues such as missing inputs, unexpected integration errors, and coding mistakes inevitably arise. A system that appears flawless to humans may still contain vulnerabilities that AI can exploit.”
While modernizing an entire enterprise system takes time, Miyakawa recommended Patching as a Service as an immediate first step, describing the solution’s AI-powered vulnerability assessments as a “vaccine for enterprise systems” that identifies weaknesses and helps build resilience against future attacks.
Miyakawa noted that SoftBank and SB OAI Japan is expanding the rollout of Patching as a Service to 3,000 eligible companies across Japan and will deploy a team of approximately 1,000 personnel to deliver the service.
“AI will bring tremendous benefits, but it also has a dark side in the form of cyberattacks,” Miyakawa said. “In today's interconnected society, we must work together to safeguard Japan's entire social infrastructure.”
Transforming the customer experience with Sierra's leading conversational AI platform
Separately, Miyakawa announced a strategic partnership with Sierra, a U.S.-based company that provides AI agent-powered customer experience platforms. As part of the agreement, SoftBank began offering Sierra's conversational AI platform in Japan as its exclusive distributor.
Key strengths of Sierra’s platform include its rapid deployability and high-quality customer interactions. In a pilot conducted at a SoftBank call center, the system achieved impressive results, with a 97% inquiry resolution rate and a 93% customer satisfaction score.
During a demonstration for the audience, an AI agent handled an inquiry about Wi-Fi service fees. It completed the entire interaction autonomously, including verifying the customer's identity, checking their contract details, recommending a more suitable plan and sending a text message for verification purposes with a link to complete the required procedures.
In addition to conducting natural voice-based conversations, the AI agent seamlessly made real-time decisions and recommendations based on customer usage history. The demonstration showed that AI agents are now taking on tasks that previously required human judgment.
Related Articles
(Posted on July 21, 2026)
by SoftBank News Editors
