This page introduces the B-031A Metaverse Development Platform Powered by AWS, which will be exhibited at the AWS Summit 2025 Day1 - Builders Fair on June 25, 2025. (Official Site)
The booth is located in the red frame area within the AWS Builders' Fair.
Table of Contents
- 1. Exhibit Overview
- 2. Cloud Gaming Technology
- 3. Platformization Initiatives
- 4. Inquiries
- 5. Survey
1. Exhibit Overview
At the event, you can experience metaverse-related solutions utilizing the latest AWS services through live demonstrations. We will also introduce the AWS architectures and technical points behind each exhibit.
1-1. CMS-Integrated Avatar System
Overview
This is an avatar system utilizing the 3D asset management CMS "VAMS".
It dynamically loads external assets and allows real-time switching between realistic and anime-style avatars.
VAMS (Visual Asset Management System)
3D avatars are centrally managed using VAMS, an open-source 3D asset management solution released by AWS. (URL)
You can register, convert, and distribute assets, and view them from a management viewer without downloading to your device.
Architecture
The demo application retrieves pre-registered avatar assets via the VAMS API.
When an asset manager updates assets on VAMS, the changes are dynamically reflected in the demo space.
Main Use Cases
- Centralized 3D asset management within companies/organizations
- Distribution of additional/updated assets after game release
1-2. Immersive Shopping
This demo uses the "Amazon Anywhere" mechanism, allowing users to purchase real Amazon products from within the metaverse space via API. *No actual payment is processed in the demo environment.
Amazon Anywhere
Amazon Anywhere enables users to search and purchase Amazon products within virtual or interactive environments such as video games, AR, and mobile apps, without interrupting their experience. (URL)
Product information, including price and shipping, can be referenced via API. With secure endpoints and Amazon login integration, users can safely use their own Amazon accounts for purchases.
Delivery and customer service after purchase are handled by Amazon as with regular purchases.
Architecture
The demo application obtains product information and processes order confirmation/orders via the Amazon Anywhere API from within the game.
The authentication process with Amazon is included. Since the game runs on AWS servers using cloud gaming technology (described later), authentication results from the user's browser are sent to the game via an external data store.
Before Amazon authentication, only general product information is available. After authentication, information such as delivery dates (depending on Prime membership) and registered payment methods is displayed, and orders can be placed.
Main Use Cases
- Seamless product sales and user purchase flow analysis within your own website
- Providing a more immersive shopping experience linked with games and the metaverse
1-3. AI Assistant
Two types of GenAI chat using Amazon Bedrock and Amazon Polly.
Different goals and roles are set via system prompts, giving each assistant a unique personality and speaking style.
Amazon Bedrock
Amazon Bedrock is a fully managed service that provides access to high-performance foundation models from leading AI companies such as AI21 Labs, Anthropic, Cohere, DeepSeek, Meta, Mistral AI, Stability AI, and Amazon via a single API. (URL)
Its main feature is providing a wide range of functions necessary for building generative AI applications based on security, privacy, and responsible AI principles. Users can easily try and evaluate various foundation models, and safely customize them with their own data using fine-tuning or RAG (Retrieval Augmented Generation).
It also enables the construction of agents that perform tasks using enterprise systems and data sources, all in a serverless environment with no infrastructure management required. Integration with existing AWS services allows for secure deployment of generative AI features into applications.
Architecture
The demo application sends user chat messages to the generative AI via the Bedrock API and returns responses. System prompts are sent along with user messages, instructing the assistant on how to respond (e.g., "You are an AWS Summit guide. Please respond in a casual manner.").
Although not used in this application, Amazon Bedrock Knowledge Bases can be used to further enhance responses by referencing arbitrary data sources placed in file servers (S3).
Responses from Bedrock are converted to speech using Amazon Polly, and characters' facial expressions are animated using lip sync technology.
Main Use Cases
- Concierge services referencing arbitrary data sources
- NPC (Non-player Character) systems with specific roles in games
1-4. Spatial Data Analytics
Player and CPU driving data from the racing game is sent to the database every second and analyzed after gameplay.
After playing, the analysis results are displayed in a new tab.
Architecture
The position and orientation data of one player vehicle and three CPU vehicles are sent every second.
Data is streamed using Amazon Data Firehose and stored in Redshift Serverless.
The analysis screen after gameplay accesses the stored data, sends the results to Bedrock, and generates feedback data.
Main Use Cases
- Designing real-world traffic flow based on user behavior analysis in the metaverse
- AI-based coaching for game play
2. Cloud Gaming Technology
The demo application uses Amazon GameLift Streams to provide browser-based game access.
Even without a high-spec gaming PC, users can access graphics-intensive games.
3. Platformization Initiatives
AWS is currently working to make it easier to develop, release, and analyze metaverse features integrated with AWS services, as experienced in this demo.
The goal is to provide features that work in conjunction with AWS services as plugins for game development tools such as Unreal Engine, making them easy to use by simply adding the plugin.
4. Inquiries
If you have any questions or concerns, please feel free to contact our booth staff.
5. Survey
After participating in the demo, we would appreciate your cooperation in completing a survey. (Takes approximately 3 minutes)
