AWS re:Invent 2025 - Quick catch up to GenAI on AWS (GBL103)

AWS re:Invent 2025 - Quick Catch Up to GenAI on AWS (GBL103)

Introduction to the Presenter

• The presenter is Shimizu, a senior manager of solutions architects who leads a team handling social infrastructure clients.
• Shimizu showcased an AI-powered self-playing piano at the recent AWS Summit in Tokyo, which generated a lot of interest.

The Evolution of Artificial Intelligence

• Within the broader framework of AI, machine learning and deep learning have been rapidly evolving.
• Generative AI uses large-scale data to train a base model with billions of parameters, which serves as the "brain" of the AI system.
• This contrasts with the traditional machine learning approach, where models are created by labeling data and learning from it, limiting their functionality to specific tasks.
• Generative AI's base model is created by learning from unlabeled data, allowing it to handle a variety of formats (text, samples, code, images, audio, video) and process information more akin to human cognition.

AWS AI Services and Offerings

• AWS AI services are defined in three layers:
  1. Infrastructure Layer: Provides the hardware and tools for building and training AI models, such as Amazon SageMaker, AWS Inferentia, and AWS Trainium.
  2. Model and Tools Layer: Offers services like Amazon Bedrock, which allows users to select and customize advanced AI models for their applications.
  3. Application Layer: Includes productivity-enhancing services like Amazon QuickSuite and AWS Transform.

Amazon Bedrock

• Amazon Bedrock is a service for building generative AI applications, offering the following key features:
  1. Model Selection: Allows users to select and use the most suitable model from a range of advanced industry models using a single API.
  2. Customization: Enables customization of models using an organization's own data to accommodate industry-specific terminology and requirements.
  3. Responsible AI: Provides guardrails to enable applications that consider security and safety.
  4. Agent Function: Supports processing of multiple consecutive tasks, enabling the handling of complex business operations.

Generative AI vs. Agent AI

• Generative AI, as discussed so far, is limited to specific tasks and still requires human intervention for exceptions and complex scenarios.
• Agent AI, on the other hand, is an evolutionary type of AI that can reduce human intervention and increase autonomy, directly impacting business operations.

AI Agents and their Business Applications

• AI agents are systems that can independently think, plan, and act to achieve goals in digital and physical environments.
• By combining generative AI models, APIs, and knowledge bases, AI agents can process tasks in a coordinated manner, adapting to changing conditions and making decisions autonomously.
• According to a Gartner report, the adoption of AI agents in enterprise software is expected to grow from less than 1% in 2024 to 33% by 2028, with 15% of daily work decisions being made independently by AI agents.

Amazon QuickSuite: Leveraging AI Agents

• Amazon QuickSuite is a service that leverages AI agents to help organizations make better business decisions by utilizing their data.
• Key features of Amazon QuickSuite include:
  1. Customizable AI Agents: Agents can be customized to suit an organization's specific needs and preferences.
  2. Quick Research and Quick Sight: Provide expert-level insights and analysis in new domains.
  3. Quick Flow and Quick Automate: Automate routine tasks and complex business processes.
  4. Spaces: Accelerate team collaboration through AI-powered features.

AI-Powered Software Development Support

• AI is transforming software development, reducing the manual effort required for tasks like code generation, debugging, and documentation.
• AWS offers several AI-powered software development support services:
  1. Amazon CodeGuru: An AI-powered service that supports efficient code generation and problem-solving through a conversational interface.
  2. Kite: An integrated development environment (IDE) equipped with agent-based AI functions, enabling usage-driven development and automated documentation.
  3. AWS Transform: A service that models legacy workloads, such as mainframes and VMware, and automatically converts them into new project versions, reducing costs and effort.

Key Takeaways and Resources

• The presenter encourages attendees to take advantage of the following resources:
  • Weekly Generative AI blog: Provides updates on the latest developments in the field.
  • Generative AI Practical Application Program: Offers support to accelerate the adoption of generative AI in businesses.
• A Japanese wrap-up session on Thursday at 2:45 PM will summarize the various service announcements made during the event.