Tokyo to Host AGNTCon + MCPCon on Agentic AI in 2026

Tokyo will welcome AGNTCon + MCPCon Japan on September 10-11, 2026, a two-day conference dedicated to the practical realities of deploying agentic AI systems. The event runs at Belle Salle Shibuya Garden, placing it in one of the city's busiest districts.

More than 40 sessions make up the program, with a stated focus on what actually happens when organizations move multi-agent systems from pilot projects into production. That framing sets it apart from broader AI conferences that lean heavily on research previews or vendor keynotes.

SoftBank and Hitachi are confirmed speakers, and both are expected to draw on internal deployment experience rather than product pitches. Their sessions will cover multi-agent trust boundaries, the mechanisms that govern how autonomous agents verify and constrain each other's actions, along with open-source agent infrastructure.

Who It's For

The agenda suggests a technical audience: engineers and architects building or evaluating agent frameworks, along with teams wrestling with the security and coordination problems that arise once multiple AI agents operate together. The Model Context Protocol angle, embedded in the MCPCon half of the event name, points to sessions on standardizing how agents access tools and data across systems.

Large Japanese enterprises headlining the speaker list gives the conference a distinct signal: this isn't a startup showcase, it's a venue where companies already running agentic systems at scale will discuss what broke and what held up.

Logistics

The conference is in-person only, held over two days at Belle Salle Shibuya Garden in Tokyo. No registration link has been published yet, so prospective attendees should watch the event's listing on dev.events for ticketing details as the date approaches. Given the scale of the program, with 40-plus sessions across two days, expect multiple concurrent tracks rather than a single-stage format.