“Multilingual AI concierge” Akihabara Guide begins service demonstration

Generative AI, translation AI, and AI avatars support tourists 24 hours a day with automatic responses and in 21 languages

 

IP DREAM Co., Ltd. will begin providing a demonstration service of a “multilingual AI concierge” Akihabara Tourism and Disaster Prevention Guide using generation AI and translation AI from January 6, 2025. This service will guide domestic and inbound tourists visiting Akihabara, Tokyo, through an AI avatar that supports 21 languages in natural conversation.

Tourists can use the service immediately by simply scanning the QR code. No prior registration, no app download, no ID or password required – just a browser.

This service is being offered as a demonstration of the “Multilingual AI Concierge” project*1, which was adopted as a Tokyo Data Platform case study project in fiscal year 2024. Using AI technology, we aim to put into practical use sustainable local information services by implementing and verifying a system for efficiently collecting and automatically learning local data, a system for automatically responding in natural language and multiple languages, and a system for improving the reliability of tourist guides in normal times and disaster prevention guides in the event of a disaster.

*1 Tokyo Metropolitan Government website: Tokyo Data Platform Case Study Project: “Multilingual AI Concierge” project developed in cooperation with the local community

https://www.digitalservice.metro.tokyo.lg.jp/busines

s/data-utilization/case-study/project-r602

■ Project promotion structure

In October 2024, we conducted a trial evaluation of the prototype service exclusively for Tokyo Data Platform members, and the service reflects the evaluation results of member companies.

■About the Multilingual AI Concierge

This service incorporates the multilingual AI concierge platform “VoiceOn® LLM,” which combines generative AI and translation AI. When a user scans the QR code and opens a browser, the language is automatically recognized and the user can start a conversation with an AI avatar in one of 21 languages. The conversation is conducted in natural language by voice.

The AI avatar refers to the learning data and responds appropriately to the intent of the question. By referring to reliable data collected from local stakeholders from the perspectives of tourism and disaster prevention, rather than the uncertain information that is overflowing on the Internet, the AI avatar can provide useful information to tourists who come to unfamiliar places.

The training data is regional data written entirely in Japanese and collected from the web, social media, open data, etc., and the AI learns from it without any processing work such as tag editing.

In order to maintain regional information services, it is essential to collect and register regional data, and we will demonstrate a method of data preparation that does not require translation work for inbound tourists or editing data for information services.

■ Service overview (demonstration)

Period: January 6th to January 31st, 2025 (open to the public)

January 6th to February 28th, 2025 (Tokyo Data Platform members only)

Location: Distributing guides with QR codes in the Akihabara area (pedestrian zones, etc.)

It can also be used outside the Akihabara area.

Available: 24 hours, free of charge

This AI refers to reliable information and answers accurately in most cases, but it may sometimes contain information that is not based on facts. Please use your own judgment when making decisions regarding your actions.

■ Contact information for service inquiries

IP DREAM VoiceOn Service and Support Center

Reception hours: Monday to Friday 9:00-17:00 *Excluding Saturdays, Sundays, and holidays

Phone number: 050-3101-3866

Email：global-product@ip-dream.co.jp

The information contained in this news release (product prices, product specifications, service details, release date, contact information, URLs, etc.) is current as of the announcement date. It may change without notice and the information may differ from the search date. Please note in advance.

This news release can also be downloaded as a PDF file from the URL below.

d78431-6-1c34161b08b85e5d01b6610860a067b2.pdf