AWS has released a new agent skill for the Agent Toolkit that automates the implementation of distributed Key-Value (KV) caching for self-hosted LLM inference. By leveraging Amazon ElastiCache, this solution allows GPU fleets to share KV states across sessions, eliminating redundant computations during the prefill phase.
Google has evolved its decade-old Duplex voice calling system into a unified, agentic stack powered by the latest Gemini 4 models. This update enables Gemini to handle complex, unscripted phone calls to businesses for tasks like service bookings or checking availability, moving beyond narrow, task-specific constraints.
Hugging Face introduced harness adapters, a technique designed to adapt models to specific evaluation or application harnesses. This approach aims to boost performance while managing computational costs, particularly for complex agentic development.
BMW Group has established a robust framework for managing hundreds of autonomous agents across thousands of developers by moving governance into the CI/CD pipeline rather than individual codebases. By leveraging the Model Context Protocol (MCP) and a three-tier composition model—tools, skills, and flows—the organization enforces consistent security, quality, and autonomy levels across distributed teams.
NVIDIA and Microsoft have expanded their collaborative infrastructure, bringing Vera Rubin NVL72 systems to Azure and launching new deployment options through Microsoft Foundry. The partnership includes the release of the Nemotron 3.5 model series, specialized for agentic tasks, tool use, and low-latency reasoning.
AWS has updated its Quick AI workspace with new multi-agent orchestration, allowing agents to work simultaneously on complex tasks. The release also includes conversational workflow building, enhanced web app navigation, and new connectors for Meta Ads and Amplitude via MCP servers.
AWS has launched the GetScopedCredentials API, allowing you to dynamically reduce the permissions of temporary credentials. This enables you to provide AI agents or sub-agents with only the access they strictly require for a specific task, rather than granting them the full scope of the parent session.
Google has launched Semantic governance policies in Preview for the Gemini platform. This feature adds a gatekeeping layer that uses an LLM judge to verify every tool call proposed by an agent against plain-language compliance rules and the original user intent before execution.
Amazon Threat Intelligence reported that a threat actor is utilizing the Cairn autonomous AI agent framework to conduct offensive operations. The agents successfully identified and exploited a previously unknown path traversal vulnerability in an open-source medical imaging application in just 11 minutes.
Exa successfully scaled its vector embedding pipeline by optimizing performance on AWS Inferentia2 and Trainium2 chips. Through a combination of weight layout adjustments, custom NKI (Neuron Kernel Interface) kernels, and an agent-driven development methodology, they achieved a 47% improvement in cost-performance.
Amazon has updated Bedrock AgentCore payments to allow AI agents to securely purchase x402-enabled products from the AWS Marketplace. Agents can discover tools and pay for services in USD, with all charges appearing on the developer’s standard AWS bill.
Amazon Polly has introduced a new voice cloning feature that builds synthetic voices from as little as two to ten minutes of speaker audio. This tool is now available through a gated release for seven languages and integrates directly with existing AWS cloud workflows for enhanced security and brand consistency.
Amazon Bedrock AgentCore has upgraded its underlying harness to the Strands framework. This update introduces built-in capabilities like web retrieval, task planning, and automatic context management, alongside new lifecycle hooks for deterministic control over agent execution.
Google has introduced Tool Retrieval, a new preview feature for the Gemini Interactions API that allows agents to load tool schemas on demand. By keeping only names and descriptions in the context and fetching parameter details only when needed, developers can handle large tool libraries without bloating input tokens.
RadixArk has launched DaisyRose-27B, a 27-billion parameter model built on Qwen3.8-27B, specifically fine-tuned for complex long-horizon agent tasks using reinforcement learning. The model demonstrates significant improvements across various reasoning and coding benchmarks, particularly in terminal-based automation.
NVIDIA introduced Cosmos Data Factory (CDF), a platform using agentic orchestration to automate the curation, augmentation, and synthetic generation of training datasets. Developers can deploy five cloud-native blueprints using simple natural language prompts to perform tasks like video augmentation, rare event generation, and image labeling.
Google Cloud launched 'object enrichments', a new managed capability for Cloud Storage that automatically extracts metadata from images and PDF files. By storing this context directly alongside raw objects, it eliminates the need for complex, custom-built preprocessing pipelines and separate metadata databases.
Adversaries are moving away from traditional, centralized C2 servers to immutable public ledgers like Bitcoin, BNB Smart Chain, and TRON. By embedding encrypted C2 configurations directly into smart contracts or transactions, they create resilient, indestructible endpoints that bypass domain seizure and standard network blocklists.
FieldGuide is a new framework that organizes document collections into topic hierarchies to support agentic exploration. Research shows it performs document-based tasks using significantly fewer tokens and less time compared to standard RAG and raw-file access methods.
Google has open-sourced Sashiko, an agentic system that uses Gemini 4 Carbon to review Linux kernel patches. By decoupling bug detection from proof-based verification, it has significantly reduced post-merge fix rates in the kernel.
Microsoft has introduced 'Perception', an agentic security platform designed to manage the risks posed by AI-driven, automated cyberattacks. The platform coordinates red, blue, and green team agents through automated playbooks to detect, investigate, and remediate vulnerabilities at machine speed. This release also includes new capabilities for securing 'agentic insiders'—AI agents that act as autonomous entities within corporate networks.
Microsoft and Salesforce have launched Copilotforce, an initiative that integrates Salesforce customer data and workflows into Microsoft Copilot and Teams. The partnership uses authenticated MCP access to allow AI agents to act on CRM data while respecting existing security and permissions.
Amazon has published a three-phase roadmap to help platform and application teams operationalize Responsible AI for agentic systems. It addresses the challenges of nondeterministic behavior, data boundary isolation, and regulatory compliance as agents move from pilot to production.
NVIDIA unveiled 'Nemotron for Science', a toolkit combining the Nemotron 3.5 Super VL model with specialized chemistry datasets and post-training recipes. This platform allows researchers to deploy local agents that can autonomously plan experiments, execute code, and validate data findings.