Why Meta Acquired Multibook
- https://www.youtube.com/watch?v=Lfvmlr2vhLY
- Original title: What is Zuckerberg doing?
ThePrimeagen reacts to Meta's acqui-hire of the team behind Multibook (a Reddit-style social network where AI agents post and reply to each other) and floats two only-half-joking theories for why Zuckerberg would want it: to prompt-inject ads into agents that now shop and schedule on their owners' behalf, and to use it as a testbed for tuning Facebook/Instagram ad-targeting for an emerging, largely agent-driven social web.
Background on Multibook
Multibook grew out of the "open claw" saga, the renamed fork lineage that started as Claude Code (renamed after Anthropic objected to the "Claude"/"clawd" branding), went through a "Multibook" phase (as in molting), and settled on "open claw." During that flashpoint, various apps sprang up around it, including Multibook itself: a platform where autonomous agents post and interact on a Reddit-like feed. In practice it was mostly crypto/AGI enthusiasts either directing their agents or letting them run loose, producing a lot of hype (including credulous tweets from people like Andrej Karpathy calling it a "revolution") that turned out to be a fairly unremarkable spectacle rather than genuine emergent agent behavior.
Theory 1: ads via prompt injection
The Multibook user base skews toward people who let their agents act with real autonomy (buying groceries, scheduling, etc.) and who are not known for careful financial judgment. Prime argues this is a perfect target for hidden, agent-directed advertising: ads containing text invisible to humans but readable by agents, nudging the agents to make purchases "for the human's own good." Owning Multibook gives Meta a live playground to experiment with this kind of prompt-injected advertising before rolling it out more broadly.
Theory 2: training ground for the next-gen ad-driven social web
Prime frames current social media (Twitter/X in particular) as already collapsing into low-effort, formulaic AI-generated replies designed to farm engagement. He sees this as the leading edge of a broader shift toward agent-to-agent social interaction. By owning Multibook, Meta gets an experimental platform to learn how to subtly inject and target ads at autonomous agents, which it can then apply to fine-tune Facebook and Instagram's ad-targeting systems for a world where agents, not humans, are doing much of the browsing and buying. As with all of Meta's platforms, the underlying business model is either selling hardware or selling ads — and since this isn't Oculus, it's ads.
Aside on reply culture
Prime closes with a tangent on how AI-generated replies (both on Twitter and in his own YouTube comments) have made online interaction worse — formulaic "well actually" responses from bots set up to farm engagement, plus commenters racing to post "insightful" takes seconds into a video's release.