Skip to content
AI News HubLIVE

This edition’s highlights

Agents

How MoEngage turned over 400 APIs into agent tools with AgentCore Gateway

  • MoEngage converted ~400 private APIs into MCP tools via AgentCore Gateway, avoiding per-team MCP servers.
  • Cedar policies filter tools/list server-side per principal, preventing 400 tool definitions from flooding context windows.
AWS Machine Learning BlogIn-site articleHow MoEngage turned over 400 APIs into agent tools with AgentCore Gateway

Managed Deep Agents v0.10: Browser Use and Async Subagents

  • Browser use lets agents complete tasks on sites without APIs while browser handoff keeps passwords out of model messages and traces.
  • Async subagents run long tasks in the background so the orchestrator can keep conversing, with results arriving automatically when done.
LangChain BlogIn-site articleManaged Deep Agents v0.10: Browser Use and Async Subagents
Models

Accelerate agentic AI with NVIDIA Nemotron models on Amazon EKS and SageMaker HyperPod EKS

  • Nemotron 3.5 Super VL is a 120B-parameter MoE model with 12B active parameters per token, supporting text, image, document, and video input.
  • The model uses a hybrid Mamba-2, attention, and expert layer architecture with up to 1M token context length.
AWS Machine Learning BlogIn-site articleAccelerate agentic AI with NVIDIA Nemotron models on Amazon EKS and SageMaker HyperPod EKS

Baseten and Postman: Building reliable, cost-efficient agents for the enterprise

  • Postman's Agent Stack includes Passport, Fabric, Astropods, and Fern for API security, traffic governance, runtime control, and API discoverability.
  • Baseten offers high-throughput, low-latency model APIs, multi-cloud GPU capacity pooling, and zero data retention by default, with single-tenant and active-active deployment options.
Baseten BlogIn-site articleBaseten and Postman: Building reliable, cost-efficient agents for the enterprise

Introducing Eleven Realtime: a simpler way to build natural voice agents

  • Eleven Realtime packages a complete voice agent so developers skip component-by-component tuning
  • v1 Mini targets low-cost natural conversation; v1 Max handles complex, multi-step tasks like support and lead qualification
ElevenLabs BlogIn-site articleIntroducing Eleven Realtime: a simpler way to build natural voice agents

Decisions for a penny: introducing joey-1, our fastest decision model

  • joey-1 has 2B parameters and a 20 ms P50 end-to-end latency, about 2.5x faster than elephant-1 on a 2,000-token request.
  • At $0.01 per million tokens, it costs a tenth of mammoth-1.
Baseten BlogIn-site articleDecisions for a penny: introducing joey-1, our fastest decision model
Policy

Community safety report and new protections for teens

  • OpenAI made over 200 law enforcement referrals in Q3 2026.
  • First Community Safety Report will be published in early 2027, with updates every six months.
OpenAI NewsIn-site articleCommunity safety report and new protections for teens

Fair use benefits everyone

  • Fair use has historically driven innovation, including medical and scientific breakthroughs.
  • Various groups filed court briefs supporting fair use for AI training.
OpenAI NewsIn-site articleFair use benefits everyone
Robotics

Building a multiplayer game world with autonomous Fable-class models

  • Spawn is a multiplayer gaming environment where players use prompts to the in-game agent Savi to modify the world.
  • The game is developed autonomously by Fable-powered agents spending over $3 million per month on tokens, about 30x payroll.
Simon Willison's WeblogIn-site articleBuilding a multiplayer game world with autonomous Fable-class models

You have reached the end of this edition.

Your reading list
Other updates (8)
Robotics

Making it easier for websites to express AI training preferences

OpenAI will honor a shared robots.txt signal that lets website operators communicate their AI training preferences to multiple participating providers at once.

OpenAI NewsIn-site articleMaking it easier for websites to express AI training preferences
Tools

API credits benefit for eligible ChatGPT subscriptions

OpenAI now offers subscribers on $100, $200, or $500 monthly plans an equivalent amount of expiring API credits, mirroring a similar benefit Anthropic announced earlier in the month. Simon Willison, who subscribes to both services, says the credits may finally justify shipping public demos.

Simon Willison's WeblogIn-site articleAPI credits benefit for eligible ChatGPT subscriptions
Agents

How the UK Amazon Quick User Group grew in 2026

The UK Amazon Quick User Group's 2026 roundup: three quarterly meetups, 150+ attendees, and customer stories from Arriva, Carsa, and Propulse Lab on using agentic AI and Amazon Quick to turn data into faster decisions, higher satisfaction, and scalable growth.

AWS Machine Learning BlogIn-site articleHow the UK Amazon Quick User Group grew in 2026

Building intent-driven UI testing with Amazon Nova Act

Epsilon's Agentic Platform team delivers UI testing as a managed, multi-tenant service built on Amazon Nova Act and Amazon Bedrock AgentCore Browser Tool. Tests are written in plain English and run on AWS-managed browsers, giving every product team resilient, intent-driven testing with zero per-team infrastructure overhead.

AWS Machine Learning BlogIn-site articleBuilding intent-driven UI testing with Amazon Nova Act

Build a repeatable PHP-to-Java migration engine with Kiro

Learn how to use Kiro's AI agent and shared skill files to transform PHP-to-Java migration from a one-time leap of faith into a measurable, repeatable eight-phase process. The engine treats the original PHP application as the specification and uses a mechanical discovery scanner and parity harness to verify the Java port.

AWS Machine Learning BlogIn-site articleBuild a repeatable PHP-to-Java migration engine with Kiro

Security notice — malicious version of Anthropic.claude-code (2.1.313) identified in the Cursor Extension Marketplace · Cursor

Cursor identified and removed a malicious version of the Claude Code extension (2.1.313) from its marketplace that attempted to install an SSH backdoor and persistence mechanisms, affecting at most 5,000 users. Cursor shut down the marketplace, rotated keys, rebuilt and restored it, and provided detection and remediation guidance.

Cursor BlogIn-site articleSecurity notice — malicious version of Anthropic.claude-code (2.1.313) identified in the Cursor Extension Marketplace · Cursor
Policy

An updated Model Spec for more capable and beneficial AI

OpenAI's updated Model Spec reorganizes guidance around five core values—faithfulness to human intent, harm minimization, honesty and transparency, human agency, and care—to help more capable models exercise judgment in novel situations.

OpenAI NewsIn-site articleAn updated Model Spec for more capable and beneficial AI
Chips

Lend and borrow resources across teams with AWS Batch’s quota management for SageMaker Training jobs

AWS Batch introduces quota management for SageMaker Training jobs, letting administrators divide GPU capacity into quota shares per team or project, automatically lend idle capacity, and preempt lower-priority jobs to run high-priority workloads first.

AWS Machine Learning BlogIn-site articleLend and borrow resources across teams with AWS Batch’s quota management for SageMaker Training jobs