Tanushree AggarwalA quick summary of Dr. Swami Sivasubramanian's keynote at AWS Re:Invent Day 4.
Dr. Swami Sivasubramanian, VP of AI and Data at AWS talked about
how the use of strong data foundation can help create innovative and differentiated customer solutions.
Customer speakers delved into how they have used data to support a variety of use cases, including generative AI, to create unique customer experiences.
Here are the key takeaways from the keynote:
Amazon SageMaker HyperPod Flexible Training Plans
- Efficiently distribute and parallelize your training workload across all accelerators.
- Save weeks of training time and help meet timelines and budgets
- Remove manual provisioning of compute capacity
- Quickly create training plan, save weeks of training
- Now Generally Available
Amazon SageMaker HyperPod task governance
- Centralized governance across all generative AI development tasks
- Optimize accelerator utilization for model training, fine tuning and inference
- Up to 40% cost reduction
- Dynamically allocate accelerator compute resources across tasks
- Ensure high priority tasks are completed on time
- Monitor and audit compute allocation and utilization in real time
- Now Generally Available
https://aws.amazon.com/about-aws/whats-new/2024/12/task-governance-amazon-sagemaker-hyperpod/
AI apps from AWS partners now available in Amazon SageMaker AI
- Enables customers to easily discover, deploy, and use best-in-class machine learning (ML) and generative AI (GenAI)
- Find, deploy and use AI apps from AWS partners within AWS SageMaker
- Use 3rd party specialized applications at various ML lifecycle stages
- Fully managed AWS experience with no infrastructure to provision or operate
- Data does not leave your AaageMaker development environment
- Support for Comet, Fiddler, Deepchecks, Lakera
- Support for more 3rd party apps coming soon
- Now Generally Available
https://aws.amazon.com/about-aws/whats-new/2024/12/amazon-sagemaker-partner-ai-apps/
Poolside coming to Amazon Bedrock early next year
- Hottest startup providing Software engineering AI for large enterprises
- Poolside assistants
malibu(tackles complex software engineering challenges) andpoint(low latency code completion predictions for developers) - coming soon to AWS! - AWS first cloud provider to provide access to Poolside
Stability AI's Stable Diffusion 3.5 coming soon to Amazon Bedrock
- Generate stunning, high-quality images from text descriptions
- Accelerate concept art creation, visual effects, prototyping and detailed product imagery
- High quality AI image generation. easily deployable art scale
- Trained on Amazon SageMaker HyperPod
Luma AI coming soon to Amazon Bedrock
- Hyper fast creation time for high quality video generation from text and images
- Experiment with fluid, cinematic and naturalistic camera motions
- Amazon Bedrock first cloud provider to get access to the latest Luma model -
Luma Ray2, a high quality text-to-image and text-to-video generation model
Announcing Amazon Bedrock Marketplace
- Access to more than 100 models from leading providers, in a single unified console experience in Amazon Bedrock
- Deploy models on managed endpoints with custom scaling policies
- Leverage Bedrock APIs to integration
- Now Generally Available
https://aws.amazon.com/about-aws/whats-new/2024/12/amazon-bedrock-marketplace-100-models-bedrock/
Amazon Bedrock supports prompt caching
- Lower response latency and decrease cost by caching infrequently used prompts
- Reduce costs by up to 90% and latency by up to 85% for supported models
- Available in Preview
https://aws.amazon.com/about-aws/whats-new/2024/12/amazon-bedrock-preview-prompt-caching/
Amazon Bedrock Intelligent Prompt Routing
- Define your cost and latency thresholds and Bedrock will route prompts to best suited model
- Available in Preview
Amazon Kendra Generative AI Index
- Out of the box vector index
- Supporting connectors for 40+ enterprise data sources (like SharePoint, OneDrive, Salesforce) with Amazon Bedrock knowledge bases
- Kendra GenAI Index supports mobility across AWS generative AI services like Amazon Bedrock Knowledge Base and Amazon Q Business,
- Available in Preview
https://aws.amazon.com/about-aws/whats-new/2024/12/genai-index-amazon-kendra/
Amazon Bedrock Knowledge Bases supports structured data retrieval
- - Fully managed RAG that supports relational queries natively
- Seamlessly integrate structured data for RAG
- Use data stored in amazon SageMaker Lakehouse, newly released S3 Tables and Redshift
- Bedrock KBs can transform natural language queries into SQL queries, allowing users to retrieve data directly from the source without the need to move or preprocess the data, reducing development time from months to days
- Improve accuracy of queries with customized content
- Available in Preview
Amazon Bedrock Knowledge Bases now supports GraphRAG
- Generate more relevant responses for gen AI apps using KBs
- Auto generation of knowledge graphs (using Amazon Neptune) to link relationships across data sources
- Enables building more comprehensive and explainable gen AI apps
- More accurate responses through a single API
- Available in Preview
https://aws.amazon.com/about-aws/whats-new/2024/12/amazon-bedrock-knowledge-bases-graphrag-preview/
Amazon Bedrock Data Automation
- GenAI ETL
- Transform unstructured multimodal data for gen AI apps and analytics
- Extract, transform and generate structured data from multi-modal content
- Generates customized outputs based on rules
- Fully managed, single API experience
- No coding required
- Prevents risks of hallucinations
- Available in Preview
https://aws.amazon.com/about-aws/whats-new/2024/12/amazon-bedrock-data-automation-available-preview/
Amazon Bedrock Guardrails Multimodal Toxicity Detection
- Configure safeguards for image content
- Enhance security of multimodal gen AI apps
- Enable consistent policy control
- Available to all models within amazon Bedrock that support image content, including fine-tuned models
- Available in Preview
Amazon Q Developer is now available in SageMaker Canvas
- Develop machine learning models in natural language, without a single line of Python code
- Q Developer will break down your objective into specific ML tasks, define problem and apply data preparation techniques on the data
- Available in Preview
Amazon Q in QuickSight Scenarios
- Run scenario analysis to help users find answers to complex problems easily
- Ask Q questions about complex business problems using natural language!
- Q automatically finds relevant data and suggests analysis
- 10x faster analysis than spreadsheets
- Accessible from any Amazon QuickSight dashboard
- Available in Preview
AWS Education literacy initiative
- Five year commitment of cloud technology and technical support for organizations creating digital learning solutions
- Empowering organizations to educate underprivileged learners globally through cloud computing
- Up to $100 million commitment of AWS cloud credits, over the next 5 years, along with support from AWS experts for training
- Deepen existing partnership with
code.orgfor providing learning platform - New partnership with
Rocket Learning- an organization supporting more than 3 million children in India
That's all for today!
Hope you enjoyed this summary! Which announcement is your favorite?
Note: No AI was used in the creation of this blog.
