🤖 AI Agents

The agents are leaving the browser — and landing on your desk.

⚡ Alibaba Cloud Drops Qwen3.7-Max and a Full Agentic AI Stack

At its first international Qwen Conference in Singapore, Alibaba Cloud launched Qwen3.7-Max (ranked 5th globally, 1st among Chinese models on the Intelligence Index) alongside Qwen Cloud, an AI-native platform with 60+ cloud products converted into MCP-compatible agent skills. The JVS Agent Suite includes JVS Claw Teams for 24/7 cloud operations and JVS Mobile for autonomous reasoning. Alibaba also joined the PyTorch Foundation as a Platinum member. For businesses in Asia-Pacific, this is a major agent infrastructure play from the region’s largest IaaS provider (22.5% market share).

📎 Newswit

⚡ Benchmark Breaks Its Own Rules: $2B Raise Signals the AI Agent Gold Rush

For two decades, Benchmark capped funds at ~$425M and stuck to early-stage. That discipline just ended. The firm closed $2 billion across two funds — a $1.25B growth vehicle (its first ever) and a $750M early-stage fund. The catalyst? A $3.25B return from Cerebras and the realization that AI agent startups need bigger checks. Benchmark recently led rounds in Gumloop ($50M, AI workflow automation) and Manus (which hit $100M ARR before a $2B Meta deal was blocked by Chinese regulators). When the VC that defined disciplined early-stage investing goes big, it’s a signal: the AI agent economy is too capital-intensive — and too lucrative — for small bets.

📎 TNW

📰 AI News

The stories shaping the AI landscape today.

Leaked: Microsoft’s Internal Strategy Aims to Make Users “Addicted” to Its AI Assistant

Documents obtained by 404 Media reveal Microsoft’s internal strategy for Project Scout (formerly Lobster/ClawPilot): explicitly “make people addicted” to its always-on AI assistant that operates across Teams, Outlook, OneDrive, and as a desktop personal assistant on both macOS and Windows. Over 1,000 Microsoft staff — including CEO Satya Nadella — are already using it. One Microsoft employee called the addiction framing “very troubling,” noting it’s language “no product should be making a part of its build strategy.” Scout targets knowledge workers in finance, legal, operations, and HR. The ethical questions are unavoidable: when “engagement” becomes “dependence,” where does user agency end?

📎 IBTimes UK

Sam Altman to Congress: Fund Testing, Don’t Require Model Approvals

Sam Altman is making a direct push on Capitol Hill, urging lawmakers to fund Commerce Department testing with cybersecurity, bioweapons, and national security scientists — but to reject any mandatory pre-release licensing regime for AI models. Altman met with House Speaker Mike Johnson, who described a “light touch” regulatory framework as the goal. The battle lines are clear: OpenAI supports evaluation and disclosure while drawing a hard line against government gatekeeping between a model and its users. The outcome will define how — and whether — the U.S. regulates frontier AI.

📎 TNW

AI Investment Boom Just Forced a Central Bank to Reverse Course

In a striking acknowledgment of AI’s macroeconomic force, Reserve Bank of Australia governor Michele Bullock told a Senate committee that the AI investment surge “supercharged” the global economy — forcing the RBA to reverse planned interest rate cuts. The bank had expected a slowdown driven by Trump-era tariff uncertainty. Instead, extraordinary AI capex ignited growth. This is the first time a major central bank has explicitly named AI investment as the reason for a monetary policy U-turn. For businesses, the message is clear: AI spending is now a macroeconomic variable that central bankers can’t ignore.

📎 Australian Financial Review