Meta (META) is riding high on its Muse AI agent. The app has hit the top spot on both Apple's App Store and Google's Play Store and, according to market intelligence firm Sensor Tower, reached 560,000 daily active users after just 11 days.
AI agents are the next step in the evolution of AI. Rather than being chatbots like ChatGPT or Claude, agents use AI models to get things done.
They aren't new, either. OpenAI (OPAI.PVT), Anthropic (ANTH.PVT), Google (GOOG, GOOGL), and Amazon (AMZN) each offer their own agents, as do open-source firms. What sets Muse apart, however, is its focus on the consumer market.
The idea is for Muse to get things done on your behalf. That includes things like firing off emails, helping organize your schedule, and shopping for you.
I took the agent for a test run to see how it stacks up to the hype and came away impressed with how well it handled most of what I threw at it, even if it ran into a few hiccups along the way.
When you first fire up Muse, you'll get the option to name the AI agent a custom name. Muse suggested a few options, but I chose to stick to plain old Muse.
And then I got to work. I started out by asking the agent to provide me with hourly updates on the latest tech headlines from a variety of news sites, something that would instantly make my work life a bit easier. A small avatar that looks like a mix between an Ewok from "Star Wars" and Finn from "Adventure Time" started tapping away on a tiny keyboard at the top of the screen, indicating that Muse was processing my request.
The agent then came back with a confirmation that it would provide the headlines and links to the appropriate sites, even noting that some required subscriptions to access the full articles. Not too shabby.
I'm in the market for a new video game for my Nintendo (NTDOY) Switch 2, so I figured, why not let Muse take the wheel and buy it for me? I asked the agent to get to it. Muse navigated to Nintendo's site and started the purchasing process, giving me a heads up as it moved through various steps. Once it got to the point where it needed to log into my Nintendo account, Muse asked for my information, noting that it would stay in secure storage and that the agent couldn't see or read what I typed.
Muse moved forward with the purchase, breaking down the total cost, saying it would charge it to my credit card on file with Nintendo, and prompting me to hit the buy button.
All in all, it was a pretty seamless experience, save for me forgetting my Nintendo login a few times.
But it was also much slower than if I simply went to the Nintendo site myself and ordered the game. I could, however, see how this would be helpful for someone who was buying a number of disparate items for, say, a party or moving supplies.
Speaking of which…I'm moving and still haven't nailed down a mover. So I asked Muse to help compare local moving companies and provide some quotes. The agent asked for my current address, where I'm moving to, the number of bedrooms in my current place, if there are any stairs or elevators, and for the email and phone number where prospective movers could reach me.
Muse found four different companies in my area and sent out my request for quotes. About two minutes later, my phone was blowing up with calls from two of the movers. One company offered a quote straight up, which Muse provided to me. The other movers sent over emails.
That's the exact kind of use case that would get me to use an AI agent in my personal life. It took the annoying task of searching for movers, and turned it into a quick back and forth before I was flooded with quotes.
I then decided I'd book a date night for my wife and me, giving Muse some basic ideas for what I was looking for. It followed up with four suggestions, one of which I ended up booking via the agent through OpenTable.
I even used Muse to help grab a new defense for my fantasy football team. I connected the agent to my Yahoo Fantasy Football account, and it searched for the defenses with the highest projected scores for the coming week. Since I could only claim them on waivers, Muse put in a waiver request for the Kansas City Chiefs and dropped Green Bay. The next day, when the claims were processed, Muse slotted Kansas City into my lineup as my active defense.
To celebrate my future fantasy win, I asked Muse to search for nearby restaurants that serve gluten-free beer. While the agent provided two correct options, it also pulled up one spot that has been closed for years. When I pointed that out, it apologized and said it initially relied on stale data.
I then asked about a casual spot that I know serves gluten-free beer. Muse said it didn't see any mention of such a beer on the restaurant's site, and suggested I call to see if it's something they have, but don't list.
Strangely, Muse provided a number but then said to use a different number instead. When I asked why the change, it said it simply made up the first phone number out of thin air and didn't confirm it. Muse then told me not to trust unverified numbers from it.
Hallucinations, or instances in which AI models confidently provide incorrect responses as if they are facts, are a known issue throughout the industry. And while companies have implemented improvements such as retrieval-augmented generation to reduce hallucinations, they still make it difficult to trust AI outputs outright.
Despite that, Muse generally proved to be a solid helper for a host of multi-step tasks.
But the AI agent will have to overcome hesitancy among some users who question whether they should share more data with Meta than they already do on Facebook and Instagram.
The company has faced criticism over the years for how it handles user information, and in 2019, it paid $5 billion to the FTC to settle claims related to privacy issues stemming from the Cambridge Analytica scandal.
More recently, Meta settled a massive social media addiction case with a group of state attorneys general, agreeing to pay upwards of $17 billion and change several business practices.
If Meta can convince users to trust Muse, though, it could prove to be a major win for the company's AI ambitions moving forward. Email Daniel Howley at dhowley@yahoofinance.com. Follow him on Twitter at @DanielHowley.