NVIDIA highlights companies at New York Climate Week using AI to accelerate clean energy projects and build a low carbon grid. ThinkLabs AI reduces grid interconnection evaluation time, Atomic Canyon optimizes nuclear operations, and Redwood Materials repurposes EV batteries for off-grid AI power solutions.
Benchling enhances security for AI agent-generated scientific code, processing 600+ sessions/day with zero incidents. Amazon Bedrock AgentCore enables multi-tenant code execution with strict data access controls.
xAI's Grok 4.6 now on Amazon Bedrock, designed for long-running agents and complex tasks. Offers configurable reasoning levels and wider surface area for Converse API and Chat Completions.
EXL's AI-powered Medical IDP solution on AWS reduces manual review of medical records by over 100 minutes per case. It combines IDP with large language models to extract, summarize, and query complex medical information, improving efficiency and accuracy in insurance claims adjudication and life underwriting.
Posit's IDE for data science, Positron, now runs on Amazon SageMaker AI, offering data access, compute, AI assistance, and collaboration capabilities. Data scientists can seamlessly work with Amazon Athena, AWS Glue Data Catalog, and Amazon S3 directly from the IDE, utilizing AI coding assistance and deploying real-time endpoints.
BMW Group uses Cloud Efficiency Analytics (CLEA) to monitor 14,000 cloud accounts, detecting anomalies in spending patterns. CLEA forecasts daily costs per account-service pair using AWS and Meta's Prophet forecasting library, alerting owners of deviations.
MIT Reads, celebrating its 10-year anniversary, shifts focus to fiction and memoirs to explore the impact of AI on society and education. President Sally Kornbluth selects Ted Chiang's "Exhalation" for fall 2026, sparking discussions on technology, humanity, and the future.
Reduce GPU waste and first-token latency by up to 82% with Amazon SageMaker HyperPod Inference Gateway, a Kubernetes-native solution that optimizes routing for large language models on GPU clusters. Real-time GPU signals ensure every inference request is placed on the best-suited pod, significantly lowering latency without requiring changes to model servers or client applications.
Machine learning regression aims to predict values accurately using metrics like accuracy, MSE, RMSE, and R2. R2 measures variance explained by the model, widely used but challenging to interpret.
Amazon Bedrock AgentCore enables developers to easily build, connect, and optimize agents securely at scale. The new AgentCore runtime offers faster responsiveness, improved memory management, and consistent startup times, catering to the evolving needs of production agents.
Neural network bistratum regressors have two hidden layers, often performing better than single-layer networks. The demo showcases a 2-3-3-1 network with tanh() and identity() activations, illustrating the input-output mechanism.
Aniimo and 007 First Light bring exciting new adventures to GeForce NOW this week. Explore Idyll with cute companions in Aniimo, and experience path tracing in 007 First Light with DLSS 4.5 enhancements.
Industrial safety AI utilizes computer vision and predictive analytics to prevent workplace dangers. Synthetic data augmentation on Amazon SageMaker AI and Amazon Rekognition improves person detection by up to 160% without hazardous photography.
MRH Trowe implements self-service AI agents for financial services employees, achieving secure access and compliance controls. The strategic use of Strands Agents, Amazon Bedrock AgentCore, and LibreChat reduces costs and streamlines processes for the German insurance broker.
AI agent at Meta saves SMEs time by capturing and sharing institutional knowledge, improving compliance and reducing organizational risk. The agent combines structured knowledge architecture with a self-improvement loop to create a permanent institutional memory for deep specialist domains.