Meta’s boss wants an AI flunky that never sleeps.
Meta Platforms’ chief executive, Mark Zuckerberg, wants everyone, inside and outside his company, to end up with their own personal AI agent.
According to The Wall Street Journal he is starting at the top, building a “CEO agent” to help him do his job, according to a person familiar with the project.
The thing is still in development, yet it is already helping Zuckerberg get answers faster, without first trudging through layers of humans.
Inside Meta, AI tools are spreading like a cold in an open-plan office, partly because using them now counts in staff performance reviews.
People familiar with the matter say the internal message board is stacked with posts from staff swapping fresh AI use cases and home-made tools.
Workers have started using personal agent tools such as My Claw, which can access chat logs and work files, then go and pester colleagues on a user’s behalf. It can even talk to other people’s personal agents.
Another internal tool called Second Brain is gathering pace, sitting somewhere between a chatbot and a full agent, according to people familiar with the matter.
Second Brain was built by a Meta employee on top of Claude and can index and query project documents, which is handy if you fancy outsourcing memory.
A Meta employee wrote: “meant to be like an AI chief of staff.”
Some staff say there is even a group on the internal board where employees’ personal agents talk to each other.
Meta has been hoovering up agent toys too, acquiring Moltbook, the social-media site for AI agents, and hiring its founders in a deal earlier this month. Meta acquired Manus, a Singapore-based startup that makes personal agents capable of executing tasks for users, and that it is using the tool internally.
The company has created a new applied AI engineering organisation tasked with accelerating the development of its large language models, with teams comprising up to 50 individual contributors reporting to a single manager.
Across the company, staff say they are being nudged into AI tutorial meetings several times a week, pushed towards AI hackathons and prodded to build their own tools to move faster.
