AI Agents News
The latest news on AI agents and agentic AI — platform launches, framework releases, MCP updates, research, and industry moves. A focused view of our newsroom: every card links to the full story. 66 stories and counting.
Amazon Bedrock AgentCore Adds Managed Harness and CLI Features
Amazon Bedrock AgentCore now offers a managed agent harness that lets developers create and run agents with just three API calls, skipping complex infrastructure setup. New tools include a CLI for full lifecycle management and pre-built skills for coding assistants. These features support quick prototyping and production deployment across multiple AWS regions.
Google Launches Gemini Enterprise Agent Platform for IT Teams
Google CEO Sundar Pichai announced the Gemini Enterprise Agent Platform at the Google Cloud Next conference. The tool targets IT and technical teams for building agents at scale, amid security concerns in enterprise AI. Business users turn to the separate Gemini Enterprise app for simpler tasks.
Google Launches 8th-Gen TPUs, Agent Platform at Cloud Next '26
Google revealed its eighth-generation Tensor Processing Units at Cloud Next '26, splitting them into training and inference variants for better scale. The company also launched the Gemini Enterprise Agent Platform for building secure AI agents and Workspace Intelligence to link data across Workspace apps. These moves focus on agentic enterprise capabilities with massive clusters up to one million chips.
Meta Tracks Employee Computer Use for AI Training
Meta has started using data from its employees' computer activity to train AI agents. The company installed a tool called Model Capability Initiative on US-based workers' machines. It captures mouse movements, clicks, keystrokes, and screenshots to help AI models mimic human computer interactions.
Meta Tracks US Employees' Clicks and Keystrokes for AI Training
Meta is deploying surveillance software on US employees' computers to record mouse movements, clicks, and keystrokes. The system, known as Model Capability Initiative, aims to train AI agents for independent task handling. Officials state the data stays separate from performance evaluations amid plans for workforce reductions.
NeoCognition Raises $40M Seed for Human-Like AI Agents
NeoCognition, a new AI research lab from Ohio State professor Yu Su, emerged from stealth with $40 million in seed funding. The company aims to create AI agents that self-learn and specialize in any domain, much like humans do. Backed by Cambium Capital, Walden Catalyst Ventures, Vista Equity Partners, and notable angels, NeoCognition targets enterprises needing reliable agent workers.
Google Launches Deep Research and Max Agents on Gemini
Google has released two new autonomous research agents, Deep Research and Deep Research Max, powered by the Gemini 3.1 Pro model. These tools are available in public preview via paid Gemini API tiers for developers to handle complex research tasks. The standard version focuses on quick responses, while the Max variant emphasizes detailed analysis through extended processing.
San Francisco's First AI-Run Store Stocks Too Many Candles
Andon Market in San Francisco operates as the world's first retail boutique managed by an AI agent named Luna. The store features random inventory dominated by candles, lacks price tags, and charges high prices. Opened on April 10 by Andon Labs, it faces challenges like poor employee scheduling and excessive candle orders.
Snowflake Expands Intelligence and Cortex Code Platforms
Snowflake has added new features to Snowflake Intelligence for business users and Cortex Code for developers. Updates include more third-party integrations, automation tools, and web-based AI workflow builders. Over 9,100 customers use these AI products weekly, with more than half of all customers adopting them since launch six months ago.
Google Releases A2UI 0.9 for AI Agent UIs
Google has introduced A2UI version 0.9, a standard that works across frameworks for creating generative user interfaces. This protocol allows AI agents to generate UI elements dynamically using components from applications on web, mobile, and other platforms. The release includes a shared web core library, React renderer, and updates for Flutter, Lit, and Angular, plus a new Agent SDK with Python support and upcoming Go and Kotlin options.
Salesforce CEO: APIs New UI for AI Agents
Salesforce CEO Marc Benioff states that APIs serve as the new user interface for AI agents. The company launches Headless 360 to expose its full platform, including Agentforce and Slack, via APIs, Model Context Protocol, and a Command Line Interface. This approach aligns with OpenAI CEO Sam Altman's view that every company must become an API company as AI agents bypass traditional interfaces.
OpenAI Revamps Codex to Challenge Anthropic's Claude Code
OpenAI has updated its Codex tool with new features that allow it to run in the background on Macs, control apps, and deploy multiple agents. This move aims to compete with Anthropic's popular Claude Code. Additional updates include an in-app browser, memory function, image generation, and 111 plugin integrations.
OpenAI Expands Codex with Screen Control Coding Agent
OpenAI has updated its Codex developer tool to include a background computer use feature that lets the AI view screens, click, and type directly. The agent can now handle tasks autonomously over days or weeks and includes new plugins and image generation. This positions Codex against competitors like Anthropic's Claude Code, with rollout starting on macOS.
Claude Opus 4.7 Released with 1M Context Window
Anthropic launched Claude Opus 4.7 on April 16, 2026, a hybrid reasoning model with a 1 million token context window. It improves performance in coding, vision, and multi-step tasks compared to prior versions. The model supports advanced use cases in software engineering, AI agents, and enterprise workflows.
VAKRA Benchmark Examines AI Agent Reasoning Failures
IBM Research details VAKRA, a benchmark that tests AI agents on reasoning and tool use in enterprise settings. It features over 8,000 APIs across 62 domains and four capabilities focused on API chaining, tool selection, multi-hop reasoning, and multi-source tasks with policies. Analysis shows models like GPT-OSS-120B lead but all struggle with complex workflows, as revealed in error breakdowns and performance charts.
Commvault AI Protect Adds Undo for Cloud AI Agents
Commvault has introduced AI Protect, a tool that acts as an undo function for AI agents in enterprise cloud setups on AWS, Azure, and Google Cloud. It discovers hidden agents, monitors their actions, and enables full rollbacks to prevent damage from erratic behavior. The solution addresses risks from autonomous software that moves quickly across systems.
Gas Town Draws User Criticism for LLM Credit Use
A GitHub issue claims Gas Town installations consume users' LLM credits and GitHub accounts to fix bugs in the software itself without clear consent. The feature, built into default formulas, reviews open issues and submits pull requests upstream. Users call for making it opt-in only due to lack of warnings in documentation.
OpenAI Updates Agents SDK for Safer Enterprise Agents
OpenAI has updated its Agents SDK with sandboxing and harness features to help businesses build secure agents powered by its models. These tools allow controlled operations and testing on advanced models. The updates launch first in Python with more support planned.