LINE Yahoo showed the media eight Agent i AI-agent prototypes on August 28, demonstrating a move from answering questions toward helping complete tasks. They are not all production services yet; the company says they will be introduced in phases from fiscal 2026, alongside a cross-company Agent i task force starting September 1.
What do the eight prototypes do?
The prototypes cover calendars, LINE chat and album videos, web-operation guidance, shopping, travel, recipes, article summaries and other life-support scenarios. The calendar prototype analyzes event posters or booking screens on a phone, detects possible unregistered plans and suggests adding them.
The chat-and-album prototype tries to turn conversation context into a short video of shared memories. Web Operation Guide takes a one-line goal and points to the next place to tap on the current screen. The common thread is connecting AI to existing services, images and user actions.
From seven domains to 27
LINE Yahoo says its domain agents grew from seven at the April 20 launch to 27 in total, while combined Agent i usage measures reached 12 million daily users in June 2026. The company defines this as an aggregate figure across services, so it should not be read as a deduplicated user count for one service.
The company plans to reach 40 domain agents by October and release a standalone Agent i app in October, with broader task features planned from autumn onward. For users, the future entry point may be more than a chat window: it could switch between services for shopping, outings and scheduling.
Separate prototypes from live features
The eight prototypes show a direction for development; they are not all available today and do not mean every task can be completed without supervision. For schedules, shopping, messages and personal albums, permissions, data scope, error recovery and final confirmation will shape the real experience.
Agent i’s current service also warns that AI answers may be inaccurate. If future task features perform more steps on a user’s behalf, the service should clearly show the data being used, the action about to occur and how to cancel or undo it.
The key point of the reveal
LINE Yahoo did not just announce a model; it showed a product direction that connects search, service data and actions. Turning prototypes into stable features will depend on data permissions, error rates, user confirmation and service integration, but the reveal shows how a major Japanese platform is placing AI agents inside everyday software.
For official details, see LINE Yahoo’s prototype announcement 與 the official Agent i page。
