Table of Contents
Welcome back to AI News Friday! 📰🤖
If you thought last week was busy, buckle up. This week we saw the “Enterprise Agent Wars” officially kick off, a billion-dollar seed round that shifted the focus from text to the physical world, and a major hardware roadmap from Meta that shows exactly where they’re placing their bets.
Here are the “Big 5” stories from March 13th, 2026, that you need to know about. 🚀
1. Microsoft Launches Copilot Cowork
Microsoft just upped the ante for the enterprise AI space with Copilot Cowork. This isn’t just a chatbot; it’s a persistent enterprise agent designed to live in your file systems and communication channels.
It can autonomously analyze project files, flag inconsistencies across different documents, and even prepare “pre-read” summaries for your next meeting based on the latest chat history and file updates.
Kenny’s Take: The key here is “persistence.” We’re moving from a paradigm where you ask AI for help to a paradigm where the AI is already working in the background, waiting for you to catch up. 🏢
2. Google Workspace: Your Drive Just Got a Brain
Not to be outdone, Google dropped a massive upgrade for Gemini for Workspace. The AI can now synthesize information across your entire Google Drive—emails, Chat, Docs, and Slides—to create entirely new deliverables.
You can now prompt Gemini to “Create a project proposal for X based on my emails with the client and the budget spreadsheet in the Finance folder,” and it will generate a multi-format draft in seconds.
3. Yann LeCun’s AMI Labs: A $1.03B Bet on World Models
In one of the largest seed rounds in history, Yann LeCun’s new startup, Advanced Machine Intelligence (AMI) Labs, secured $1.03 billion in funding. Backed by Nvidia and Bezos Expeditions, AMI Labs is moving away from the “next-token prediction” of traditional LLMs.
Instead, they are focusing on “world models”—AI that understands the physical laws of reality. Their goal? Solving the intelligence gap in robotics and automated manufacturing.
4. Meta’s Hardware Roadmap: The MTIA Era
Meta isn’t just building models; they’re building the silicon to run them. They announced a roadmap for four new generations of their MTIA (Meta Training and Inference Accelerator) chips.
While the MTIA 300 is already in production, they are fast-tracking the 400 through 500 series to specialize in generative AI inference. This is a clear move to reduce their reliance on external hardware and optimize for their specific Llama-based workloads.
5. Nvidia’s NemoClaw: Open-Sourcing the Agent Stack
Nvidia is making sure they stay at the center of the agent ecosystem with NemoClaw. This open-source platform is designed to help software companies deploy agentic models across any hardware configuration.
By open-sourcing the training and deployment stack, Nvidia is effectively trying to set the industry standard for how “agentic” software should be built and run.
🔍 Tool of the Week: Actian AI Analyst
If you’re dealing with massive business glossaries, this is for you. Actian AI Analyst uses a governed Semantic Knowledge Graph to turn complex business data into a conversational interface. No more SQL queries to find out your MRR trends—just ask.
⚡ Quick Hits
- Anthropic’s Enterprise Marketplace: You can now buy third-party Claude-powered apps directly within the Anthropic ecosystem.
- Cribl’s Agentic Telemetry: A new way to use AI for security and IT operations, moving from “searching logs” to “asking agents.”
- LinkedIn’s Rise in Citations: AI chatbots are now citing LinkedIn twice as often for professional queries, proving the value of community-driven knowledge.
What’s your take? Are you ready to let an AI “Coworker” into your files, or does the rapid rise of World Models for robotics feel like we’re finally entering the age of I, Robot? Let me know! ✌️
— Kenny