Meta didn’t buy Moltbook for bots — it bought into the agentic web
When news broke Tuesday morning that Meta had purchased Moltbook, the social network for AI agents, it may have left some people scratching their heads. What on earth would Meta – an ad-supported company – want with a social network whose users are bots? After all, bots are not the target group of brand marketers and advertisers.
Meta doesn’t say much. The only official comment was a brief statement that the Moltbook team was joining Meta Superintelligence Labs, which would “open new ways for AI agents to work with people and companies.”
Read between the lines: this was an acqui-hire. A network built for bots isn’t exactly a natural home for brand advertising, even if Moltbook never quite was non-human. What Meta really wanted was the talent behind it: people who have fun brainstorming and experimenting with ecosystems of AI agents. And that, counterintuitively, could be a boon for its advertising business.
Like Meta CEO Mark Zuckerberg said last yearhe believes in a future where “every company will soon have a business AI, just like they have an email address, social media account and website.” On an agent web, where AI systems act independently on behalf of users, AI agents can communicate with each other and do things like buy ads, make bookings, and respond to customers.
AI is too is used to generate creative and tailor its output to who is viewing it. AI systems can also manage product prices or generate personalized offers.
On the consumer side, agents can be used to find the best prices and offers, manage bookings and purchase products. In a limited number of cases, agents can already settle and pay on behalf of the consumer. (Agentic trading is still in its infancy and these systems do not always work as well as advertised. But the market moves quickly and improvements seem likely soon enough.)
Just as Facebook once built the “friend graph”—a network defined by social connections between people, where each individual is a node—an agentic web could benefit from an “agent graph,” a system that maps how different agents are connected and what actions they can take on each other’s behalf.
However, for an agentic web in which corporate agents and consumer agents can work together, the agents must first be able to find each other, connect, and coordinate their activities. Just as Facebook once built the “friend graph”—a network defined by social connections between people, where each individual is a node—an agentic web could benefit from an “agent graph,” a system that maps how different agents are connected and what actions they can take on each other’s behalf. This can cover areas such as travel, online shopping, media and research, productivity tools and more.
Advertising could also play a role here. Today, people view and click on ads when they see something interesting, but on an agent web where agents shop on behalf of users, ads can look very different. Instead of influencing a human to buy a product, a company’s agent may have to negotiate directly with a consumer’s agent to make the sale.
Maybe the consumer wants to buy that shirt or lipstick, but only in a certain color and at a certain price. Perhaps the systems are becoming so complex that these considerations go beyond product and price; perhaps consumers prefer to support small businesses, or shop only at environmentally friendly companies. Perhaps the consumer only buys items when they are on sale, or buys generic versions when the ingredients are the same. And so on.
In that case, it’s not just a matter of connecting the AI agents, but also of ranking products based on what best suits the needs of that individual customer. If Meta could capitalize on that market — AI at the orchestration layer, meaning the system decides which agents talk to each other and in what order — it could potentially expand its advertising business into entirely new territory.
This all depends on whether consumers actually embrace the agent web, or whether they ever trust AI enough to let it act on their behalf. But the very existence of Open Clawthe AI personal assistant that has fueled Moltbook suggests that at least some people are already leaning on autonomous AI agents.
Of course, there is another possible reason why Meta bought Moltbook. The company lost its acquisition to OpenClaw creator Peter Steinberger to compete with OpenAI, so it went after Moltbook, the platform that helped build Steinberger’s tool. Petty? Maybe. But it kept Meta’s Superintelligence Labs in the news.
