Sky Cone

[14:32] [New] Amazon S3 Express One Zone

Highest performance and lowest latency cloud object storage

- Co-locate data and compute
- Millions of access requests per minute
- Single-digit millisecond latency
- 50% lower access costs vs S3 Standard


[18:30] [New] AWS Graviton4

The most powerful and energy-efficient chip we have ever built

- 30% faster than Graviton3
- 40% faster for database applications
- 45% faster for large Java applications


[18:57] [New] R8g Instances for EC2
Powered by AWS Graviton4


[35:48] [New] Amazon EC2 Capacity Blocks for ML

Reserve EC2 UltraClusters to run your ML workloads

- Reserve EC2 UltraClusters with hundreds of GPUs
- Ideal for training & fine-tuning FMs, short-duration workloads & handling capacity surges


[38:20] [New] AWS Trainium2

Purpose-built chip for generative Al and ML training

- Optimized for training FMs with hundreds of billions to trillions of parameters
- 4x faster than AWS Trainium
- 65 exaflops of on-demand supercomputing performance


[42:00] Amazon Bedrock

The easiest way to build and scale generative Al applications with LLMs and other FMs

- Choice of industry-leading FMs from AI21 Labs, Amazon, Anthropic, Cohere, Meta, and Stability Al
- Customize FMs using your organization's data
- Enterprise-grade security and privacy

[ - 55:40] Anxxx


[] Amazon Bedrock customization capabilities

[New] Fine tuning

Cohere Command Lite, Meta Llama 2, Amazon Titan Text Lite & Express

GENERALLY AVAILABLE

Anthropic Claude
COMING SOON

[59:10]
[New] Retrieval Augmented Generation (RAG) with Knowledge Bases
GENERALLY AVAILABLE

[1:00:21] [New] Continued Pre-training for Amazon Titan Text Lite & Express
AVAILABLE IN PREVIEW TODAY


[1:00:00] [New] Agents for Amazon Bedrock
Execute multi-step tasks across company systems and data sources

- Enables generative Al applications to take action in just a few clicks
- Breaks down and orchestrates tasks and executes API calls on your behalf
- Securely accesses and retrieves company data

GENERALLY AVAILABLE


[1:07:44] [New] Guardrails for Amazon Bedrock
Safeguard your generative AI applications with your responsible AI policies

- Easily configure harmful content filtering based on your responsible Al policies
- Apply Guardrails to any FM or agent
- Redact PII information in FM responses (coming soon)

AVAILABLE IN PREVIEW TODAY

[1:09:25 - ?] Pfizer

[]
Amazon CodeWhisperer customization capability

- Better and more relevant code suggestions
- Your data is never used to train the underlying model
- Onboard developers faster


[1:28:07] [New] Amazon Q

A generative Al-powered assistant for work that is tailored to your business

AVAILABLE IN PREVIEW TODAY

- Provides interactive answers, solves problems, generates content, and takes action
- Understands your company information, code, and systems
- Personalizes interactions based on your role and permissions
- Built to be secure and private


[1:29:24] Amazon Q

Your expert assistant for building on AWS

- Trained on 17 years of AWS knowledge
- Assists you everywhere you work with AWS - in the console, IDE, and documentation
- Converses with you to explore new AWS capabilities, learn unfamiliar technologies, and architect solutions
- Works with you to troubleshoot, build new features, and upgrade languages on AWS

To make Amazon Q appear in the Visual Studio Code, install or update the "AWS Toolkit - Amazon Q, CodeWhisperer, and more" extension.



[1:35:14] Amazon Q Code Transformation

- Complete Java language upgrades in a fraction of the time
- Enhance security posture and performance
- Accelerate migration from Windows to Linux (coming soon)

Currently (Dec. 7, 2023), Amazon Q can upgrade Java 8 and Java 11 code to Java 17 code.


[1:37:17] Amazon Q

Your business expert

- Delivers quick, accurate, and relevant answers to your business questions, securely and privately
- Connects to over 40 popular data sources including S3, Salesforce, Google Drive, Microsoft 365, ServiceNow, Gmail, Slack, Atlassian, and Zendesk
- Respects existing access controls - only returns info you're authorized to see based on your role


[1:38:21 - 1:41:35] Amazon Q demo


[1:44:36] Amazon Q in Amazon QuickSight

- Generative dashboard authoring
- Visually compelling data stories
- Reimagined Q&A experience

[1:45:21] Amazon Connect
Contact center in the cloud

- Set up a contact center in minutes and easily scale to support millions of customers
- Use built-in Al and ML to automate and personalize customer interactions
- Cost savings up to 80% compared to traditional contact center solutions


[1:46:14] Amazon Q in Amazon Connect

- Generative Al contact center agent assistance
- Real-time customer conversations generate recommend agent responses
- Natural language search for generated responses and links to more information
- On-board agents faster with less training


[01:49:00 - 1:57:08] BMW

[2:03:00] Zero-ETL integrations with Amazon Redshift
[NEW] Amazon Aurora PostgreSQL
[NEW] Amazon RDS for MySQL
[NEW] Amazon DynamoDB

AVAILABLE IN PREVIEW TODAY


[2:05:02] Amazon OpenSearch Service

Securely unlocks real-time search, monitoring, and analysis of operational data

- Uncover insights faster with personalized search & analytics
- Proactively detect performance and security threats
- Tens of thousands of active customers


[2:05:11] [New] Amazon DynamoDB zero-ETL integration with Amazon OpenSearch Service

GENERALLY AVAILABLE


[2:06:00] Amazon DataZone

Catalog, discover, share & govern data across your organization

- Automatically populates a data catalog to improve data findability
- Search and discover data
- Create data projects to bring your datasets and analytics tools together


[2:11:00] Amazon Project Kuiper

-
-
今年AWS在Re:Invent 2023大会上发布的foundational和AI类的service会有哪些？



-