Table of Contents
A CNET reporter recently tested Meta Muse to streamline the process of finding second-hand household items on Facebook Marketplace. By setting up automated search parameters, the AI agent helped track down specific products, monitor pricing, and initiate contact with sellers.
Setting Up the AI Agent
To begin, the reporter accessed the Marketplace prompt through the “Ideas” sidebar in the Meta Muse interface. Unlike a standard one-time search, this feature functions as a recurring task. The user provided specific criteria, including a list of desired items—an ottoman, an air purifier, and a humidifier—along with preferred locations and price points. For example, the reporter set a budget of $150 for the ottoman and $50 each for the air purifier and humidifier, specifying a preference for Montclair, New Jersey, with the option to travel to Manhattan for specific items.
Once configured, Muse began scanning Facebook Marketplace every four hours for new listings that matched these requirements. The tool also provided helpful context, such as identifying the retail value of specific brands and flagging potential concerns, such as when a seller was new to the platform.
Managing Communications and Searches
When the reporter decided to purchase an air purifier, Muse drafted a message for the seller. After the user reviewed and edited the draft, the AI agent connected to Messenger to send the inquiry. Muse continued to monitor the conversation, checking for seller responses every 30 minutes and providing notifications to the user.
The process proved effective for more subjective searches as well. When looking for a midcentury-style ottoman, the reporter used the chat function to refine results based on specific design preferences. The AI allowed the user to ask for more information on specific products and even expanded the search to include other items, such as espresso machines and non-toxic air fryers.
The Verdict on Marketplace Automation
According to the reporter, using Muse offered a more dynamic experience than manually scrolling through endless Marketplace listings. The ability to have the AI run in the background and automatically search for alternatives if a product sold or a seller failed to respond was noted as a significant benefit.
While the AI agent is capable of handling initial communications, the reporter preferred to take over conversations manually once a seller replied to ensure the tone remained personal and to avoid sharing sensitive information like phone numbers or addresses. Although the agent’s 30-minute check-in interval meant the user sometimes saw replies before the AI notified them, the tool successfully helped the reporter make progress on their shopping list, ultimately leading to a completed transaction for one of the items.
