AWS Launches Various AI Innovations in the Latest Re:Invent, What Are These?
AWS (Amazon Web Services) have organized their annual event, AWS Re:Invent 2024, in Las Vegas, United States. Not long ago, AWS Indonesia brought various innovations from the event to Jakarta, including technology-based Artificial Intelligence (AI) which is the main highlight.
Focus on AI Innovation: Efficient and Reliable
In AWS Indonesia’s presentation, AI technology is at the core of the innovations offered. AWS claims that these innovations can provide high performance with lower operational costs, and support various organizations in Indonesia to make more optimal use of AI technology.
“We have deliberately designed this innovation so that it has a real impact on the way our consumer’s business works and the growth of the business,” explained Paul Chen, head of the solution architecture for the ASEAN region at Amazon Web Services.
AWS divides their innovations into four main categories:
AI technology advancement
paradigm shift in work method
Infrastructure and hardware development
data as the foundation of AI
Amazon Nova and Bedrock: Answering the Needs of Various Organizations
What is Amazon Nova?
Amazon Nova is the latest AI model set developed by AWS. This is the latest version of Foundation Model high-performance, available through the service Amazon Bedrock. Amazon Nova provides the organization with flexibility to choose the AI model that best suits their needs.
Models in Amazon Nova include:
Amazon Nova Micro
Amazon Nova Lite
Amazon Nova Pro
Amazon Nova Canvas
Amazon Nova Reel
Amazon Nova Sonic
Each model has a different specialization:
Nova Micro, Lite, Pro: Understanding Model (text, image, video as input → output text)
Nova Canvas, Reel: Creative Content Generation Model (input text/image → image output/video)
Nova Sonic: Speech-to-Speech Model (voice input → sound & text output)
To date, Nova Canvas, Nova Reel, and Nova Sonic models only support English. Meanwhile, Nova Micro, Lite, and Pro support more than 20 languages, according to AWS.
Full Service at Amazon Bedrock
Amazon Bedrock is an AWS service that allows companies to build generative AI applications more easily, thanks to access to various Foundation Model From AI21 Labs, Anthropic, Cohere, Meta, Mistral AI, Stability AI, and Amazon itself through one unified API.
AWS calls Amazon Bedrock the most practical method of developing generative AI applications, with an emphasis on security, data privacy, and ethical use.
New models and features in Bedrock
Two Foundation Model The latest introduced:
Luma Ray2 (by Luma AI): Text input → video output
Stable Diffusion 3.5 large (by Stability AI): Text input → image output
Bedrock also presents Marketplace with more than 100 Foundation Model, including those from Asia Pacific.
Another excellent feature is Automated Reasoning Checks which helps verify the accuracy of generative AI output, while explaining how AI answers are formed — as part of GENERATIVE AI SAFEGUARDS belongs to AWS.
Amazon Q Developer: AI Assistant for Developers
AWS also introduced updates to Amazon Q Developer, a developer-specific AI assistant. New features include:
Modernization of Windows .NET Framework applications into Linux versions
Transformation of VMware workload to architecture Cloud-native
These features are claimed to be able to speed up and simplify the digital transformation process.
AWS Trainium: Generative AI Chip
The AWS Trainium chip line is specially designed for generative AI model training. latest version, Trainium2, offers four times the performance and power efficiency twice as much as the previous generation. AWS also introduces Trainium3, which promises higher performance.
AWS has launched Amazon EC2 trn2 which is powered by Trainium2 and is now generally available.
Amazon DynamoDB: Consistent Global Replication
Amazon DynamoDB now support Multi-Region Strong Consistency on Global Tables. This feature ensures that global applications can access the latest versions of data in various AWS regions, very useful for organizations that require low latency and high consistency.
Case Study: Halodoc Use Amazon Nova Pro
In Indonesia, Halodoc become one of the companies that have adopted the latest AWS technology, in particular Amazon Nova Pro. As a digital health ecosystem platform, Halodoc processes various claims from more than 3,000 partner hospitals.
With Nova Pro, Halodoc implements Zero-Touch Auto-Adjudication for:
Extracting information from claim documents
grouping benefits
Test data accuracy
Prevent fraud
“The AI approach of Nova Pro enables significant efficiency of claim processing costs, so that the process can be more Scalable and economical,” he said Dody Alfian, Senior Engineering Manager, Information Security at Halodoc.
The implementation of this AI successfully cut claim processing time by 50% and increased work efficiency by more than 45%, with a decrease in AI costs by about 50% compared to similar models.
