Over the past few days, Zuckerberg finally had a moment to revel in his success.

Not only did his company’s stock soar by 11.3%, but he also managed to give Tencent a 5-point boost.

What sparked this surge? Muse. Within just 12 days of its launch, Muse has amassed nearly 2.8 million total installs, reached 642,000 daily active users in the US mobile market, and catapulted to the top of the US App Store’s free chart—even surpassing ChatGPT in downloads during the same period.

For many, their initial reaction upon seeing Muse was: “Didn’t China already try this AI assistant thing?”

Earlier this year, nearly every major Chinese tech company jumped into the personal AI assistant race. There was AI for food delivery, AI for ride-hailing—demos were everywhere. For a while, it seemed like super apps were about to be reimagined by AI agents.

But then? The excitement faded quickly.

Yet here we are, six months later, and Americans have taken the baton, turning it into Silicon Valley’s hottest new story.

Today, AI personal assistants have swiftly become one of the most talked-about trends in Silicon Valley.

Beyond Muse, a wave of AI personal assistant products has emerged, including Instinct, Town, and Grok Bot. The most remarkable part? Instinct hasn’t even officially launched yet, but its valuation has already hit $2.5 billion.

At this rate, AI personal assistants might even overtake AI tools for office productivity.

It’s surreal. The same personal AI assistant concept failed to gain traction in China but suddenly exploded in the US. Why? Let’s dive in.

/ 01 / Eliminating Friction Costs: Muse’s Core Value

Many assume that once personal agents become mainstream, people will primarily use them for AI-powered shopping, dining reservations, and trip planning.

You’d be mistaken.

What do Americans love using Muse for the most? Bargaining. Or more precisely, saving money.

Foreign research firm FUNDA analyzed 654 public use cases of Muse over its first 12 days. The top category? Saving money, securing refunds, and managing personal finances—128 cases, accounting for 19.6%.

Here are a few standout examples:

Someone had Muse call AT&T to negotiate their $85/month, 500Mbps broadband plan down to $40/month for 1Gbps;

Another user slashed their Verizon Fios bill from $134 to $64;

One user waited 2.5 hours for AT&T customer service without resolution—Muse dialed two lines simultaneously and connected to a human in 10 minutes;

The most clever case? A sports fan had Muse automatically activate and cancel different streaming services each month based on the NFL schedule, cutting their season costs from $310 to $181.

This thing isn’t just an AI assistant—it’s a digital bargainer.

FUNDA calculated that across 49 cases with clear monetary savings, users saved a combined $35,426 annualized—$723 per case on average, with a median of $490.

Beyond cost-saving, administrative tasks and travel are also top use cases for Muse.

After reclassifying all cases, FUNDA found that 62.8% of Muse’s use cases fall under “Life Admin”—handling life’s hassles for users.

These scenarios share two traits: First, they’re tedious but rarely urgent, so people procrastinate. Second, the results translate directly into dollars saved, making the value crystal clear.

In other words, Muse’s main value lies in eliminating friction costs in consumer scenarios.

But why does this work in the US and not China?

The answer is simple: The friction costs of daily life in America are astronomically high.

An average person’s life is scattered across dozens of disconnected systems—flight info in emails, itineraries in Google Calendar, car insurance on insurer websites, credit cards in banking apps, broadband plans in carrier portals, medical records in hospital-specific patient portals.

Even more absurdly, many services lack full digital closed loops. You can buy insurance online, but canceling it might require a call. You can order broadband online, but getting a discount often demands human intervention.

By 2026, some US financial processes will still rely on fax machines.

When Business Insider tested Instinct, they encountered a quintessentially American scenario: To get a bank to waive a late fee, Instinct had to fax documents on the journalist’s behalf.

Some processes still require faxes. During Business Insider’s Instinct testing, to secure a late fee refund, Instinct ended up faxing documents to the bank.

Now compare this to China, where super apps have already eliminated most of the incremental value of personal agents. Booking hotels, paying bills, scheduling doctor appointments, processing returns—all can be done with a few taps.

In other words, the $2 friction costs Americans struggle with were already smoothed out in China years ago.

/ 02 / Riding the Open Ecosystem Wave The second reason foreign personal assistants took off? Openness.

Muse’s logic resembles Doubao Mobile, but with a key difference: Instead of operating your phone, Muse gives each Agent its own cloud-based computer.

Meta creates a Muse Secure VM for each user, where the AI interacts with your accounts and data. Muse then browses the web, searches, fills forms, shops, and sends emails independently.

For Muse to work, Europe and America’s internet ecosystem needed to already be third-party-friendly.

Take email: Gmail long ago provided full APIs and OAuth authorization. With one click, third parties can read emails, search info, and even act on your behalf—within clear permissions.

So when Muse needs to check your flight time, it doesn’t ask you—it just scans your inbox for tickets and hotel confirmations.

This openness defines much of Europe and America’s internet. Standardized interfaces dominate entire service chains.

For example: Expedia exposes hotel search, availability, pricing, and booking APIs; Shopify’s Storefront API lets third parties read products, build carts, and proceed to checkout; Stripe offers standardized payment processing.

Crucially, even without APIs, the web remains intact.

US airlines, hotels, e-commerce sites, banks, carriers, and government services almost all maintain full web interfaces. For today’s Computer Use-capable Agents, this is enough. If APIs fail, they can navigate webpages, click buttons, fill forms, and contact customer service just like humans.

Muse’s success, then, is partly due to the open ecosystem legacy of the past two decades in Europe and America.

/ 03 / Don’t Overestimate Muse After Muse’s explosion, investment banks immediately started hyping it up.

Morgan Stanley just published an “agentic” playbook, crunching numbers for Muse: 100 million monthly active users, 5 queries per user per day, 10% with commercial value, $0.07 per commercial query, 365 days a year—equals ~$1.3 billion in annual revenue and ~1% EPS growth by 2028.

That breaks down to $13 per MAU annually. At an 80% incremental profit margin, that leaves just $2.60 per user per year to cover new service costs at the same scale.

The math looks great. But here’s the problem:

Saving money is a fantastic story for viral growth, but not necessarily for user retention.

Muse’s rave reviews all come from saving hundreds on car insurance, broadband bills, or flight cancellations—clear, tangible savings that users love to share.

But these are one-time wins, harvesting existing errors, negotiable bills, and forgotten subscriptions. Once exhausted, sustainable high-frequency scenarios are scarce.

More critically, Muse struggles to penetrate higher-frequency core scenarios due to supply-side resistance.

Already, Amazon has started restricting Muse’s access.

This makes perfect sense. What Muse is doing is essentially rebuilding a consumption gateway: Users tell Muse what they want instead of searching/comparing/ordering on Amazon themselves. Amazon becomes a background supplier, reduced to merely providing products and fulfillment.

That’s untenable for Amazon, whose value lies not just in inventory and logistics but in its user relationships and search entry points. No company wants to be a replaceable supplier hidden behind someone else’s interface.

Here’s a bold prediction: America’s cherished open ecosystem will face reevaluation in the Agent era and may gradually converge toward China’s walled gardens.

“All things divide and reunite.” This ancient wisdom holds true everywhere.

Ultimately, the consumer market is a supplier-driven market. Demand-side optimizations are inherently fragile.

Muse’s gateway narrative is far less rosy than it appears.

By Qi