Edition #7. The week of NVIDIA GTC. Three developments that together show where AI agents now stand.
MiMo-V2: the first model built for the Agent era
One of the engineers behind DeepSeek R1 launched MiMo-V2. Not just an update, but a complete model family specifically designed for AI agents. Pro, Omni and TTS versions. Built on a 1-trillion-parameter base model with a 1 million token context window.
This is exactly what's been observed in recent months. The models from a year ago were good at having conversations. The models of today are good at executing tasks. That's a fundamental difference. ChatGPT was a conversation partner. MiMo-V2 is a coworker.
The model is optimized for complex agentic workflows where an agent combines multiple tools, sources and steps to complete a task. Ultra-low latency, 1M context window, and it goes open source as soon as it's stable enough.
The data confirms this. On OpenRouter, the platform where developers compare and use AI models, Hunter Alpha, the model behind MiMo-V2, sits at number 1. 1.51 trillion tokens processed. A growth of 2,017 percent this week. On that same platform OpenClaw sits at number 1 among apps with 699 billion tokens, four times more than number 2.
NVIDIA launches NemoClaw: enterprise AI agents become the standard
Jensen Huang stood on stage for three hours on Monday at GTC 2026 in San Jose. His message was clear: every software company in the world needs to have an OpenClaw strategy. He called OpenClaw the new computer and said it has done in a few weeks what Linux took 30 years to do.
The biggest announcement: NemoClaw. An enterprise-grade security and privacy layer on top of OpenClaw. Local AI models on your own hardware, policy-based guardrails on agent behavior, and a privacy router that only switches in cloud models when needed. One command install.
NVIDIA also unveiled Vera Rubin: 7 new chips specifically designed for agentic AI, with 10x more performance per watt than Blackwell. NVIDIA expects $1 trillion in orders through 2027.
China builds OpenClaw clones
The lobster has escaped the pot, headlined the South China Morning Post this week. The New York Times ran a long story on OpenClaw in China. Long lines in Shenzhen of people seeking help installing it. Local governments offer subsidies, free compute and discounts on office space to companies using OpenClaw.
But the Chinese government also warned about security risks. Chinese tech companies are jumping on it en masse. Minimax launched MaxClaw, Zhipu released a similar tool. Stocks are rising on investor optimism about AI agents.
What I did with AI agents this week
This week, Atlas, my own OpenClaw agent, processed the full administration of two BVs and researched dozens of potential customers. That's an AI agent. Not a chatbot. A digital employee that executes tasks.
The real power sits in the combination with custom software. Fully configured agents on dedicated servers. Not a bare install, but working digital employees with their own tools. Think of a sales agent with its own CRM that finds leads for you daily, analyses websites, and writes personalized emails. Or a bookkeeping agent that's available 24/7, proactively processes your receipts, calculates VAT, and keeps your administration up to date before you even think of it.
No sick days. No vacations. No salary negotiations. Always available, always consistent, and a fraction of the cost of an employee.
