Meta's New AI Agent Muse Wants to Live in Your Phone

Meta used its big annual Connect event to show off Muse, a new AI agent that runs inside your phone. An "agent" is AI that doesn't just chat — it takes actions, like looking things up or handling small tasks for you. While rivals OpenAI and Anthropic chase business customers who pay big money, Meta is going the opposite direction: straight at regular people, the same way Facebook, Instagram and WhatsApp already sit in billions of pockets. Meta also showed a cute, Tamagotchi-style AI gadget that it insists is for adults only.

TechCrunch reporter Sean O'Kane tried Muse himself. The most impressive moment: it actually found him money he was owed and never claimed. But his overall verdict was that it feels like "a party trick" — fun the first time, not something you'd open every day. That matters, because an AI assistant only becomes useful if it earns a spot in your daily routine. A neat demo isn't the same as a habit.

The bigger question is trust. Meta's core business is advertising, and the more it knows about you, the more it can charge advertisers. As Sean put it, Meta will argue that knowing you better means better ads — but plenty of people aren't buying that yet. Handing an always-listening, always-watching AI your messages, photos and shopping habits is a very different deal from handing them to a company that sells software subscriptions.

So is this smart or a misstep? Meta has one huge advantage: it already knows how to be part of ordinary life in a way business-focused AI companies do not. If anyone can make AI feel normal rather than nerdy, it's Meta. But normal only works if people trust it — and that is the one thing Meta has struggled to earn for years.