Picture a woman I know who runs a small catering business out of her home kitchen. She lives in her phone — voice notes to clients, photos of pastry trays, invoices scribbled between delivery windows. Her WhatsApp is basically her storefront, her calendar, and her filing cabinet rolled into one. Last weekend, over coffee, she showed me something on her screen with a mix of delight and suspicion: an AI assistant living right inside her chat app, offering to handle her bookings, her reminders, her quotes. “Is this the part where it starts doing my job for me,” she asked, “or the part where it starts billing me?”
Both, it turns out. Possibly this week.
Last week Meta launched Muse, its consumer AI agent, in what The Daily Upside described as one of the most consequential product launches of the year. The details are striking. Muse comes in a free tier plus paid subscriptions at $20 and $100 a month — a pricing ladder that puts it squarely in competition with the premium chatbots people already know. It is available in a standalone app and, more importantly, inside WhatsApp, which counts 100 million monthly active users in the United States alone, according to The Daily Upside. And it is US-only for now, which means this is a controlled experiment with one of the largest captive audiences in consumer tech.
Wall Street has noticed. Morgan Stanley analyst Brian Nowak put out a client note arguing that Meta could win a sizable piece of a $30 trillion consumer AI pie, and he suggested the share price could gain 25 percent by New Year’s Eve. That is the kind of forecast that makes portfolio managers sit up: a thirty-trillion-dollar opportunity, even sliced thinly, changes the valuation math for a company that already dominates the attention economy.
But here is where the story gets more human — and more honest. Reuters reported, via The Daily Upside, that some of Meta’s own employees see Muse as unreliable, and that security flaws have been flagged. That matters, because an AI agent is not a chatbot you ask for a poem. It is a thing you hand your accounts to — your reservations, your calendars, your money. Meanwhile, Business Insider reported that Resy, the restaurant reservation platform, is banning users who hand their accounts over to AI agents. Think about that: the agent economy has barely started and already one of its most natural uses — having your AI book you a table — is running headfirst into a platform that does not want machine customers.
There is a bigger shadow over all of this, too. Axios reported, via Muck Rack, that Tim Cook is handing the Apple reins to John Ternus after fifteen years at the top. Apple closed Monday at $316.61, compared with $13.35 when Cook took over in August 2011 — a 2,272 percent gain. Cook’s tenure is a case study in what a successor inherits: a mountain of compounding returns, a fanbase of investors, and the impossible question of whether the next chapter can rhyme with the last. Every big-tech transition story is really a question about whether the company’s future product — whatever it is — can earn the valuation the market has already priced in.
It’s not enough to just download the newest app and marvel at what it can do. It’s not enough to treat a 25 percent share-price forecast as a tip rather than a thesis. We must listen to what the builders themselves are saying about reliability, learn how these tools actually fit into — or disrupt — the platforms we use every day, and contribute our own judgment instead of outsourcing it. Because the most expensive mistake in a technology transition is not missing the wave. It is boarding the wrong one without looking at the hull.
My take: the $30 trillion number is the headline, but the Resy story is the tell. If restaurants are already banning AI agents from their booking systems, then the friction in the agent economy is real and arriving early. Meta’s bet is that 100 million WhatsApp users will meet Muse where they already live — in their conversations. That distribution advantage is enormous. But distribution without trust is just a loud megaphone. The employee doubts reported by Reuters are a flashing yellow light: an agent that books the wrong flight or mishandles your money does not get a second chance from most people. I am bullish on agents as a category and cautious about every specific product until it has survived contact with real customers.
For regular people, here is the practical version. First, do not pay $100 a month for an AI agent you have not tried for free. The free tier exists; let it earn the upgrade. Second, be thoughtful about what you hand to any agent — your restaurant logins, your calendar, your payment details — the way you would be thoughtful about handing your house keys to a new assistant. Check the fine print on platforms you care about: some, like Resy, are already drawing lines. Third, if you own Meta stock or an index fund that holds it, understand that the bullish case now depends on a consumer AI bet landing in the real world, not just in a note. A 25 percent gain by New Year’s Eve is a forecast, not a promise — and forecasts are cheapest thing on Wall Street.
My friend with the catering business ended up trying the free tier. She told me it drafted a quote letter for a wedding client in about forty seconds, and she spent twenty minutes fixing the tone. “It saves me time,” she said, “but I’m still the one signing my name.” That might be the truest sentence anyone has said about AI agents all year. The tool is real. The judgment is still ours. And the companies that remember which one the customer is actually buying — the capability or the trust — are the ones that will still be standing when the forecasts fade.
