Summary
Japanese startup MetAI says it will debut a locally operated AI tourism guide at NIIGATA AI EXPO on October 28–29. The prototype runs speech recognition, retrieval, response generation, voice synthesis and lip movement on one HP workstation, without sending conversations to the cloud. MetAI says operators can build a virtual human in under an hour from a short recording or choose anime, 2D, 3D and full-body characters. It currently detects and answers in Japanese, English and Spanish, using only registered official information and directing unknown questions to a contact rather than guessing. The company plans a public pilot at NINNO3 in Niigata from December 2026 through February 2027, working with Ask, HP Japan and local partners.
Hana’s perspective
The interesting part is not the talking avatar by itself, but the choice to run the whole interaction locally and limit answers to approved material. That combination can reduce privacy risk and the confident invention of tourist facts. For animation and character businesses, it also offers a route from a mascot’s visual design to an interactive installation without depending on constant connectivity. The claims still come from MetAI’s press release, so the expo demonstration and later pilot will need to show that multilingual latency, lip sync and refusal behavior remain convincing outside internal tests.
Are you a writer with a great story? We'd love to hear from you: APPLY HERE