Monday just launched Agentalent.ai, a managed marketplace where enterprises can post roles and hire autonomous AI agents for functions like marketing and operations. Built with AWS and Anthropic, the platform signals a major shift toward blended human-AI workforces.

Co-founder Roy Mann believes every company will soon employ both humans and AI agents. Pricing starts at $2,000 per month for specialized agents, and early partners like Wix are already evaluating roles for digital workers to handle campaign execution and ops. Rather than treating agents as background integrations, the platform mirrors traditional talent acquisition with a structured hiring process.

Agents undergo authentication and capability checks before deployment to ensure they meet rigorous enterprise security standards. This move signals that software vendors are becoming AI employment agencies for the modern era. While the initial roster is small, it marks a move beyond experimentation toward operational integration of autonomous agents.

Anthropic has acknowledged that users of its Claude Code tool are hitting usage limits much faster than expected. The company is investigating the issue as a top priority after developers reported exhausting monthly quotas in just hours of coding. The disruption affects both Pro and Max plan subscribers.

Some frustrated users on the $200/month Pro plan claim they only get twelve usable days out of thirty. The root cause may be a bug inflating token usage via prompt cache failures, highlighting the scaling challenges AI providers face as demand for autonomous coding tools surges.

Anthropic recently tightened Claude Code usage limits without notifying users, sparking widespread developer frustration. As AI models take on more complex tasks, balancing heavy automation with user expectations remains a moving target. Some users have found downgrading versions helps, but the core issue remains under active investigation.

🛡️ Security researchers discover DNS data exfiltration flaw in ChatGPT

🤝 Stanford study finds AI chatbots are prone to sycophancy

🧠 Meta FAIR team launches TRIBE v2 to predict brain activity

⚖️ Governor Newsom signs order tightening AI rules for California

🎭 South Korean actress likeness used without permission in AI film

🎙️ Alibaba releases Qwen3.5-Omni with real-time voice cloning features

Here's what to do

📂 Step 1: Pick a task you keep putting off Something you'd normally spend 20+ minutes on. Comparing flights, researching restaurants for a group dinner, or pulling together competitor info for work.

✨ Step 2: Activate Agent Mode Open ChatGPT and select "Agent Mode" from the tools menu or type /agent. Be specific with your prompt. The more context, the better.

🚶 Step 3: Walk away. Tasks typically take 5 to 30 minutes. Go make coffee or knock out something else. You'll get a notification when it's done.

🔍 Step 4: Review and refine Check the results when you're back. You can jump in and steer if needed. Agent Mode pauses before any high-impact actions, so nothing irreversible happens without you.

🔁 Step 5: Set it on repeat. If it's a recurring task, click the Clock icon and schedule it daily, weekly, or monthly. Manage everything at chatgpt.com/schedules.

Will your next coworker be a piece of software from a marketplace? And can AI companies keep up with the massive compute demands of power users? See you next time! 🚀

Stay informed, stay curious, and stay ahead with Jumble!

Zoe from Jumble