AI news, product launches and company developments.
OpenAI & AnthropicAmazon SageMaker AI now integrates with open source Evidently to monitor discriminative ML models. This solution helps maintain prediction accuracy by generating monitoring reports and detecting data drift. It also organizes results in MLflow and supports scalable pipelines with automated notifications.
Amazon Bedrock AgentCore enables building an AI-powered AWS Support Companion. This agent integrates with AWS services to analyze logs, search documentation, and manage support cases via a conversational interface. The solution deploys easily with AWS CloudFormation and includes a web frontend for user interaction.
Amazon Bedrock now supports MiniMax models, enabling scalable on-demand inference for various workloads. These models help build agentic applications, long-context document analysis, and software engineering workflows. The service offers multiple tiers and APIs with AWS security and operational guarantees.
Amazon SageMaker AI now integrates with MLflow to stream benchmark and recommendation results automatically. This allows real-time tracking of metrics, parameters, and charts in a unified interface. The integration simplifies experiment management and improves monitoring efficiency.
Amazon Bedrock uses AI to detect sophisticated AI-generated phishing emails. These phishing attempts are increasingly advanced due to generative AI and open source intelligence. This technology helps improve email security by identifying unique, AI-crafted phishing messages.
Amazon SageMaker AI shares best practices for multi-turn reinforcement learning training. The guidance includes building reliable training environments, setting up external evaluations, and designing task-aligned rewards. These practices help improve agent performance and monitoring during multi-turn interactions.
Amazon Bedrock now supports NVIDIA Nemotron and OpenAI GPT OSS models in AWS GovCloud (US). This enables US-based users to run large open-weight AI models with data residency compliance. The service offers multiple tiers and inference options for government and regulated industries.
AWS demonstrates HippoRAG, a neurobiologically inspired retrieval-augmented generation model, using its cloud services. The implementation leverages Amazon Bedrock for LLMs, Amazon Neptune for graph databases, and Personalized PageRank for advanced graph analytics. This showcases how to build scalable HippoRAG applications on AWS infrastructure.
Inscribe developed an AI system using Amazon Bedrock to detect document fraud quickly. The system analyzes financial documents like an expert fraud analyst. It detects tampered and AI-generated documents in under 90 seconds, improving speed by 20 times over manual review.
Amazon Bedrock introduced the open source Model Profiler to simplify model selection by aggregating metadata from various AWS APIs and external sources. It offers a single, searchable interface to help users compare and choose models efficiently. The tool can be deployed quickly, supporting practical real-world use cases.
AWS demonstrates deploying BoltzGen on SageMaker AI for protein design. The setup supports quick validation and production batch processing. It uses step-level caching to reduce compute costs during iterative research workflows.
Anthropic has launched Claude Sonnet 5, its most advanced Sonnet model, on AWS platforms including Amazon Bedrock and Claude Platform. This new model offers improved intelligence for coding, autonomous agents, and professional tasks. It provides high performance at competitive Sonnet pricing, making it accessible for various applications.
