Bedrock Brief 24 Dec 2025
Well, well, well. Looks like Amazon's been busy playing musical chairs with their AI leadership. Peter DeSantis, the AWS veteran who's been around since the digital stone age (or at least 2006), is now taking the reins of a shiny new AI-focused organization. Sorry, Rohit Prasad—looks like you're getting the "thanks for playing" consolation prize as you head for the exit.
But wait, there's more! Amazon's AI assistant, Alexa+, is getting some fancy new friends. Angi, Expedia, Square, and Yelp are joining the party, promising to turn your living room into a one-stop-shop for everything from hotel bookings to salon appointments. Because who doesn't want their AI assistant to judge their travel plans and haircut choices?
Meanwhile, the AI job market is looking about as stable as a Jenga tower in an earthquake. Over 50,000 layoffs were blamed on AI in 2025, with tech giants leading the charge. But let's be real—is AI really the culprit, or just a convenient scapegoat for companies realizing they went a bit overboard during the pandemic hiring spree? Either way, it seems the future of work involves a lot more algorithms and a lot fewer humans. Time to start buttering up your robot overlords, folks.
Fresh Cut
- NVIDIA's Nemotron 3 Nano, a powerful language model with a 256k token context window and native tool calling, is now available on Amazon Bedrock, enabling developers to build advanced AI applications with improved reasoning and efficiency. Read announcement →
- Deploy MiniMax-M2, a powerful 230 billion parameter AI model optimized for coding and agent tasks, quickly and easily through Amazon SageMaker JumpStart's interface or Python SDK. Read announcement →
- AWS End User Messaging now offers AI-powered phone number registration reviews, helping developers submit accurate and complete forms to mobile carriers on the first try. Read announcement →
- AWS Neuron SDK 2.27.0 brings native PyTorch support and enhanced tools for machine learning on Trainium3 UltraServer, making it easier for developers to optimize AI workloads without changing their existing code. Read announcement →
- SageMaker Studio reduces container startup times by 30-50% using SOCI indexing, allowing data scientists to quickly access custom environments for faster ML development and experimentation. Read announcement →
- Amazon Bedrock Data Automation now refines field extraction instructions using a few labeled documents, improving accuracy for tasks like invoice processing without model training. Read announcement →
- AWS now lets you use account-level tags for cost management across multiple accounts, simplifying cost analysis and budgeting for organizations with complex account structures. Read announcement →
- Amazon EC2's R8i and R8i-flex instances, featuring custom Intel Xeon 6 processors with 2.5x more memory bandwidth, are now available in Asia Pacific and South America, offering up to 60% faster performance for specific workloads like NGINX web applications. Read announcement →
- Lambda durable functions, which let you build reliable multi-step apps and AI workflows with checkpointing and automatic recovery, are now available in 14 more AWS regions, enabling developers to create complex processes closer to their users and data. Read announcement →
- Vercel users can now instantly create and connect to AWS databases like Aurora and DynamoDB directly from the Vercel Marketplace, with $100 in credits to get started and serverless options for cost-efficient scaling. Read announcement →
The Quarry
Move Beyond Chain-of-Thought with Chain-of-Draft on Amazon Bedrock
Chain-of-Draft (CoD) is the new kid on the block in AI reasoning, promising to outperform Chain-of-Thought by mimicking how humans actually tackle problems. Instead of long-winded explanations, CoD encourages models to use quick, punchy thinking steps that cut straight to the chase. For the tech-savvy crowd, this approach not only speeds up inference time but also shows improved performance on complex reasoning tasks, especially when implemented with Amazon Bedrock's foundation models. Read blog →
More posts:
- Exploring the zero operator access design of Mantle
- AWS AI League: Model customization and agentic showdown
- Accelerate Enterprise AI Development using Weights & Biases and Amazon Bedrock AgentCore
- How dLocal automated compliance reviews using Amazon Quick Automate
- Advancing ADHD diagnosis: How Qbtech built a mobile AI assessment Model Using Amazon SageMaker AI
- Accelerating your marketing ideation with generative AI – Part 1: From idea to generation with the Amazon Nova foundation models
- Introducing Visa Intelligent Commerce on AWS: Enabling agentic commerce with Amazon Bedrock AgentCore
- Move Beyond Chain-of-Thought with Chain-of-Draft on Amazon Bedrock
- Deploy Mistral AI’s Voxtral on Amazon SageMaker AI
- Enhance document analytics with Strands AI Agents for the GenAI IDP Accelerator
- Build a multimodal generative AI assistant for root cause diagnosis in predictive maintenance using Amazon Bedrock
- Introducing SOCI indexing for Amazon SageMaker Studio: Faster container startup times for AI/ML workloads
- Build and deploy scalable AI agents with NVIDIA NeMo, Amazon Bedrock AgentCore, and Strands Agents
- Bi-directional streaming for real-time agent interactions now available in Amazon Bedrock AgentCore Runtime
- Tracking and managing assets used in AI development with Amazon SageMaker AI
- Track machine learning experiments with MLflow on Amazon SageMaker using Snowflake integration
Core Sample
Agentic AI Transformation: Workforce Strategy & Leadership
Agentic AI isn't just another IT buzzword—it's a cross-functional revolution that's redefining competitive advantage. AWS experts argue that your secret weapon isn't the AI tech itself, but your proprietary data, which needs to be liberated from silos and made accessible across the organization. They emphasize the importance of building integrated teams and fostering a culture of genuine experimentation, where leaders actually reward calculated risks instead of just paying lip service to innovation. Watch video →
More videos:
- AWS EDGE Innovation Spotlight: Cloudsmith
- AWS EDGE Innovation Spotlight: DataDome
- JFrog, NVIDIA, and AWS: Building responsible and secure AI at scale
- AWS Innovation with PhysicsX | Innovation Ambassadors
- Tehanu: Rewarding communities protecting the world's most endangered species