An Analyst Tested Meta’s Muse for 10 Days — Then Sold Airbnb and Bought More Meta

October 3 (IT Home) — Appearing on the finance podcast The Synopsis on 29 September, veteran independent stock analyst Mostly Borrowed Ideas (MBI for short) said: after downloading and trying Muse, I sold out of Airbnb and added to Meta.

IT Home notes: MBI is the pen name of independent analyst Abdullah Al Rezwan, who runs the investment research site and Substack newsletter MBI Deep Dives, known for in-depth research on listed companies across industries, investment journals and public discussion of his holdings.

In the recent interview, MBI discussed Meta’s Muse AI agent in depth, saying that after testing Muse for about 10 days, he concluded that AI agents could bypass platforms such as Airbnb and complete bookings directly, so he sold his large Airbnb position and added to Meta.

In the interview, MBI noted that while using Muse, he found the AI agent opens the Airbnb website like a real person, enters the destination and number of guests, scrolls through search results and then makes recommendations.

Muse also supports automatic price comparison and bypassing platforms. After MBI asked Muse to find a farmhouse to stay in, Muse found that booking directly with the host was about 60% cheaper than going through Airbnb; it had already linked a payment method and was waiting only for a single confirmation from the user to complete the booking. MBI believes AI agents may weaken the control that online travel agencies, food delivery and e-commerce platforms have over the “traffic gateway” and “transaction matching.”

Muse can read a user’s past Airbnb stays and reviews, and combine them with travel Reels they have saved or watched on Instagram, to recommend places within a two-hour drive of home.

A single platform such as Airbnb struggles to offer this kind of “life context data,” whereas Muse can connect data across platforms and data types, providing personalised recommendations genuinely based on a user’s data and preferences.

On the podcast, MBI put forward the concept of “proactive commerce.” Traditional shopping falls into two categories: search-and-buy, where the intent is clear, and discovery buying while scrolling a feed.

AI agents, by contrast, may proactively alert users to cheaper or more suitable alternatives, and even place orders on their behalf. Muse, for example, once told a user that the software they were using had a much cheaper alternative with similar features.

Users also do not have to repeat their preferences every time — they can simply tell Muse once: for example, “quality first, then price, then delivery speed.” After that, the AI will follow those preferences when shopping, booking accommodation or ordering food, which significantly reduces decision friction.

On the podcast, however, MBI also pointed out that Muse is not perfect at this stage: the browser-style operation is slow — comparing just five hotels took 14 minutes — and early results may be text only, lacking pictures, reviews and maps. So it is currently more like the prototype of a future model; but the guest believes it may be the worst version there will ever be, and will only keep improving.

He therefore believes AI is turning from a “chat tool that answers questions” into a personal agent that can browse, compare, recommend and complete transactions on a user’s behalf — and may therefore reshape the business models of e-commerce, travel and food delivery platforms.

Yesterday (2 October), MBI published an in-depth opinion piece on his MBI Deep Dives newsletter titled “Why Muse may never need advertising,” arguing that AI agents such as Muse have begun to disrupt the platform economy.

MBI argues that Meta’s AI assistant Muse may not need to show ads inside the product at all, and could still create advertising value for the company. He argues that Muse’s usage and browsing activity may supply incremental data to Meta’s advertising system — especially its ad optimisation and user profiling — indirectly boosting ad revenue on Facebook and Instagram.

Muse’s conversations, searches and browsing behaviour can help Meta better understand users’ interests and purchase intent, and improve ad targeting on other platforms. Even if the Muse interface itself carries no ads, it may still “earn money” by strengthening the effectiveness of Meta’s wider advertising system.