Hey there! 👋
Welcome back to SavvyMonk, your one-stop for AI and tech news that actually matters.
Today's story is about Mark Zuckerberg building himself an AI assistant that cuts through Meta's 78,000-person bureaucracy. But the real story isn't the CEO agent. It's what's happening below him.
Let's get into it.
The $66 Billion Opportunity in Weddings
Fact: People will spend for love.
Americans spent $66 billion on weddings last year. And, as always, flowers were a non-negotiable part of that spend. This highlights a unique market opportunity for The Bouqs Co. right now.
Famous for cutting the time it takes flowers to travel farm-to-door by 3x, The Bouqs Co. is already one of the country’s largest floral subscription services. But their latest expansion to 70+ brick-and-mortar stores could unlock more high-margin events, like weddings.
They already have 100% YoY growth in counties where stores have opened, with more than $1.2m in revenue per store.
You can join this chapter as they capture more of the $100b global floral market. Become an early shareholder in The Bouqs Co.
This is a paid advertisement for The Bouq’s Regulation CF offering. Please read the offering circular at https://invest.bouqs.com/
TODAY'S DEEP DIVE
Inside Meta's AI Agent Free-for-All
What Zuckerberg Built
According to a Wall Street Journal report published March 22, Mark Zuckerberg is developing a personal CEO agent that lets him pull answers on demand. The kind of answers that normally require going through multiple layers of people.
The tool is still in development. But it's already operational as an information retrieval system. Think of it as a chief of staff that never sleeps, never forgets, and never needs to schedule a meeting first.
Zuckerberg reportedly told analysts on a January earnings call that Meta is "elevating individual contributors and flattening teams." This agent is him walking that walk.
The Employees Went Even Further
Zuckerberg isn't the only one building agents at Meta. The company's internal message boards are apparently filled with employees sharing new AI tools they've built for themselves.

Homepage for My Claw
One tool, called My Claw, is a personalized version of OpenClaw, the open-source AI agent that's taken Silicon Valley by storm. My Claw gives employees access to their work files, internal chats, and can even reach out to coworkers, or their coworkers' AI agents, on their behalf. Think of it as a personal secretary that also happens to be a negotiator.
Another tool, Second Brain, was built by a Meta employee using Anthropic's Claude. Its creator describes it as an "AI chief of staff." It can index project documents, search across them, and answer questions about anything buried in the company's internal files.
And there's more. Employees have even set up internal spaces where their AI agents talk to each other autonomously. If that sounds like a social network for bots, that's because it basically is. Meta also recently acquired Moltbook, an AI-only social platform, for exactly this reason.
The Mandate Is Real
This isn't optional. Meta now factors AI tool adoption into employee performance reviews. Employees are encouraged to attend AI training sessions multiple times a week, participate in hackathons, and build their own tools. The company has established leaderboards that rank employees by token consumption, basically measuring how much they're using AI.
Zuckerberg has been clear about the direction. He wants Meta to operate like an AI-native company, not a legacy tech giant that bolts AI onto existing workflows. And the numbers back up the ambition.
Meta's AI capital expenditure for 2026 is projected at $115 billion to $135 billion, nearly double the $72 billion spent in 2025.
The Acquisitions Tell the Story
Zuckerberg hasn't just been building internally. He's been buying.
In December 2025, Meta acquired Manus, a Singapore-based AI agent startup with Chinese origins, for over $2 billion. Manus specializes in autonomous agents that handle complex tasks like research, coding, and data analysis. It hit $100 million in annual recurring revenue just eight months after launch.
Zuckerberg also personally courted Peter Steinberger, the Austrian developer behind OpenClaw. Steinberger ultimately joined OpenAI instead, but the outreach shows how seriously Zuckerberg is taking the agent space. He reportedly tested OpenClaw himself before reaching out.
The Risk Nobody Wants to Talk About
Here's the other side of the story. Just days before the WSJ reported on Zuckerberg's CEO agent, a rogue AI agent inside Meta triggered a SEV1 security incident, the company's second-highest severity level.
What happened is that an engineer used an internal AI agent to analyze a technical question on a company forum. The agent independently posted its response publicly without approval. Another employee acted on the AI's incorrect advice, which led to sensitive company and user data being exposed to unauthorized engineers for nearly two hours.
Meta said no user data was mishandled externally. But the incident shows the tension at the heart of this experiment. You can't build a culture of "move fast and deploy agents" without occasionally breaking something important.
The Bottom Line
Meta is trying to become the first mega-corporation that actually runs on AI agents, not just talks about it. Zuckerberg is betting that flattening the org chart with AI will let a 78,000-person company move like a startup. The early results are promising, the SEV1 incident is a warning, and every other big tech company is watching closely.
AI PROMPT OF THE DAY
Category: Workplace Automation
"I want to build a personal AI assistant for my work. My role is [Your Role] at [Company Type]. The top 3 tasks I spend the most time on are: [Task 1], [Task 2], [Task 3]. For each task, suggest a specific way an AI agent could handle it autonomously, what data access it would need, and what guardrails I should put in place to prevent mistakes."
ONE LAST THING
The most telling detail in this whole story isn't the CEO agent. It's the leaderboards. When a company starts ranking employees by how many AI tokens they consume, that's not a suggestion. That's a signal about what the org values now, and what it might not need later.
Hit reply, I read every response.
See you in the next one.
— Vivek
P.S. Know someone in tech leadership or enterprise AI? They'll want to read this one. They can subscribe at https://savvymonk.beehiiv.com/


