After years of trying to convince the world he’s not a robot (and drinks water like a normal human), Mark Zuckerberg is finally steering into the skid.
Driving the news: Meta’s head honcho is building his own personal AI agent to serve as a sort of second CEO, per The Wall Street Journal. The agent’s still being built, but it’s already helping Zuck collect information from across the company faster and communicate with employees on his behalf.
It’s all part of Meta’s internal AI push, with employees now building their own agents, tools, and even message boards for their AI counterparts to interact with each other. The company also now factors how much staff use AI into their performance reviews.
Catch-up: Many of these autonomous tools are built on OpenClaw, which is a framework for operating continuously running AI agents. These individual “Claws” can access your files, talk to people, and automatically carry out tasks on your behalf.
You can create individual Claws for specific work. At Meta, employees are using a tool called My Claw that has access to emails, documents, and can autonomously reach out to co-workers (or their AI agents).
Why it matters: These personal AI agents are being pitched as the future of the workplace, freeing up time for workers, saving money for businesses, and increasing overall productivity. As Jensen Huang put it at a Nvidia conference last week, “Every single company needs an OpenClaw strategy."
Major AI players like Anthropic, Nvidia, and Perplexity are either building their own services on OpenClaw or designing alternative platforms that do roughly the same thing.
Yes, but: Trusting these AI agents to act autonomously comes with its risks. Anecdotes abound of users accidentally wiping their hard drives or inboxes, and one of Meta’s in-house agents recently went rogue and posted advice to a company forum without approval from an employee. That advice, which one worker acted on, ended up compromising sensitive company and customer data.—LA
