Generative AI on AWS: Tools on AWS that make it a preferred choice for building GenerativeAI apps

HIGHLIGHTS:

1. EC2 instances allow you to use NVIDIA H100 GPUs now
2. Amazon Bedrock now allows use of more foundational models
3. Generative business intelligence on AWS for prompt based dashboard creation
4. Training resources

 

 

AWS believes that their services and capabilities will democratize the use of Generative AI now that it has reached its tipping point. In a recent event, Amazon introduced a slew of new features available or soon to be available on AWS. All of these are directly geared towards making AWS the preferred choice for developers building or learning how to build generative AI models, apps and more.

 

Some use cases for Bedrock based AI apps include:

• Chatbots
• Search
• Image generation
• Text summarization

 

Here’s a little detail about what is possible with AWS in terms of generative AI today:

1. Model choice: AWS is expanding Amazon Bedrock to include Coherence as an FM (Foundational Model) provider. Coherence is planning to add “Command”, its flagship text generation model along with “Coherence Embed”, its multi-lingual text understanding model.Anthropic’s “Claude 2” is also now a part of Amazon Bedrocks list of language models available.Stability AI has announced the addition of their latest Stable diffusion model to Bedrock as well – for superior image detail, realism in the generated graphics. Know more here.
2. Agents: Foundational models usually require additional programming to perform more complex tasks apart from summarization. Bedrock now enables creation of agents to automate complex tasks, customization, and up-to-date answers generated using proprietary data. With just a few clicks, agents for Amazon Bedrock will automatically break down tasks and create an orchestration plan—without any manual coding, making the task of programming generative AI applications easier for developers.
3. Amazon QuickSight now combines its ML innovations with new LLM capabilities using Amazon Bedrocks updated list of FMs – giving it generative capabilities and hence allowing generative Business Intelligence on AWS.This capability will help break down silos, speeding up collaboration across the enterprise and allowing for quicker data-driven decisions. Text based prompts to generate dashboards will become a reality for business users.
4. HIPAA eligible service – HealthScribe will enable AI to be used securely in a clinical setting, enhancing productivity and simplifying compliance.
5. The ever-popular Amazon EC2 compute cloud now harness the power of NVIDIA H100 GPUs that are optimized for training LLMs and developing generative AI applications. This makes AWS the first major cloud provider to democratize access to the highly sought after H100 GPUs in production.
6. Training for free: In an effort to enable adoption of AI, AWS has released on-demand trainings for generative AI, including training and developing Foundational Models, executive trainings on AI and partner guidance trainings.

Summarizing:

AWS offers a wide range of generative AI capabilities, including:

• Foundation models: These are large, pre-trained models that can be used to generate text, images, and other content.
• Tools and services: AWS has a number of tools and services that make it easy to build and deploy generative AI applications. These include Amazon SageMaker JumpStart, Amazon Lex, and Amazon Polly.
• Developer resources: AWS offers a number of developer resources to help you learn about and use generative AI on AWS. These include blog posts, tutorials, and documentation.