The AI agent race hit full sprint this week. Meta confirmed an internally built "Hatch" agent modeled on OpenClaw, targeting completion by end of June — the same week Google's secret "Remy" 24/7 personal agent was reported by Business Insider. Nvidia and ServiceNow jointly launched Project Arc, an enterprise OpenClaw-style autonomous agent, while Microsoft declared AI agents "the next operating layer" for all Microsoft 365 work. On the product front, xAI's Grok expanded with connectors to Gmail, GitHub, Notion, and Google Calendar — making it the first Grok model with real personal context. BlackRock CEO Larry Fink raised Wall Street's antenna by declaring compute futures will become a new asset class, HKUST researchers unveiled AI agents with permanent self-improving memory requiring zero retraining, and — in the week's sleeper story — Google Chrome quietly pushed a 4GB Gemini Nano model to millions of users without any permission prompt.
Meta is training an internal AI agent codenamed "Hatch," inspired by OpenClaw, with the goal of completing internal testing by end of June. The agentic assistant is designed for regular users and is part of Meta's broader push into personal AI as competition from OpenAI and Google intensifies.
Nvidia (NVDA) and ServiceNow (NOW) jointly unveiled Project Arc, an enterprise AI agent platform similar to OpenClaw built for business workflow automation. The partnership signals that the top AI hardware maker is now moving aggressively into the agent software stack.
Microsoft unveiled its agent-first strategy on May 5, positioning AI agents as the next operating layer for all Microsoft 365 and Agent 365 work. The announcement marks Microsoft's most direct pivot away from copilot-assistance toward fully autonomous AI task execution.
Anthropic published a formal commitment that Claude will remain ad-free — citing incompatible incentives between advertising and genuinely helpful AI. The same week, Claude Code shipped a broad update with smarter model picking, improved permission handling, Windows/PowerShell fixes, and stronger telemetry.
May 2026's agentic AI security roundup highlights the Claude Code source leak, Mythos autonomous exploit demos, and emerging agent containment frameworks. As agent autonomy scales, security researchers warn the attack surface is growing faster than the tooling to contain it.
03 / In their own feeds
Ten voices on X.
Selected posts from the edition, with links to the originals.
Cointelegraph · @Cointelegraph01 / X
"🔥 UPDATE: xAI's Grok now supports connectors for Gmail, GitHub, Notion, Google Calendar, Drive, and more."
(The Wolf Of All Streets): "WATCH: BLACKROCK CEO LARRY FINK SAYS 'I ACTUALLY BELIEVE A NEW ASSET CLASS WILL BE BUYING FUTURES OF COMPUTE, WE JUST DON'T HAVE ENOUGH COMPUTE POWER RIGHT NOW.'"
"Google Chrome just downloaded a 4 GB AI model to millions of computers. No permission asked. The file is called weights.bin. It is Gemini Nano. It sits in your browser data folder under OptGuideOnDeviceModel. Most users have no idea it is there."
(Aakash Gupta): "Nvidia just figured out how to put an AI data center on the side of your house. And pay you to host it. Each XFRA node packs 16 Blackwell RTX Pro 6000 GPUs, 4 AMD EPYC CPUs, and 3TB of RAM in a Dell PowerEdge rack mounted next to the AC condenser. The homeowner pays nothing for the hardware."
"🚨BREAKING: HKUST just gave AI agents permanent memory that improves over time. No retraining required. Lessons from one task carry over to the next automatically."
(Charly Wargnier) (RT): "think Hermes is just another agent? think again. this thing is a BEAST 💥 — what actually runs under the hood: Persistent memory, multi-agent orchestration, task continuity across sessions."
(God of Prompt): "Most ideas don't disappear because they're bad. They disappear because the moment passes. TRAE SOLO Mobile helps solve that. Capture an idea on your phone → turn it into tasks → continue execution on desktop. The gap between thinking and doing just got smaller."