KDDI's Hojo: AI agent adoption stuck at 7.1％ without "AI-ready" systems

KDDI frames the problem as one of infrastructure, not AI capability. Advanced AI technology, it argues, produces no value until it is woven into frontline operations, and today that weaving has barely started: autonomous AI agents that work across multiple systems have reached only 7.1% adoption.

The cause, in KDDI's account, is how companies have approached system migration. Many pushed workloads to the cloud while leaving applications and data structures as they were, which leaves AI agents unable to reach the data they need. Closing that gap means consolidating siloed data and modernizing so that AI can read and write safely through standard interfaces such as APIs and the Model Context Protocol. Hiroki Hojo, who heads KDDI's AI integration business, calls that state "AI-ready" and describes it as the infrastructure condition that decides whether AI adoption succeeds.

KDDI positions itself around that gap, offering support that runs from building secure foundations to on-site deployment of AI-equipped robots. Whether this becomes a real dividing line for enterprises is likely to depend on how quickly legacy data and applications can be modernized — and on whether standard interfaces such as APIs and MCP are enough to make agents trustworthy with core business systems.