An analysis said Meta's personal AI agent Muse may find it difficult to generate big profits from shopping transaction fees alone. The reason is that even if an AI finds products and completes payments on behalf of consumers, the number of transactions that can realistically be delegated is limited.

According to IT outlet Business Insider on Sept. 28, analysts at MoffettNathanson said it is unlikely that personal AI agents will disrupt e-commerce as much as some investors expect. Meta's Muse and startup Instinct offer functions that search for products online and make purchases on users' behalf. Meta CEO Mark Zuckerberg has mentioned a revenue model of taking small fees from such transactions.

The key is the number of payments an AI can take over. Based on data from the Federal Reserve Bank of Atlanta, MoffettNathanson assumed consumers make an average of 47 payments a month. Excluding transactions where there is little reason to use an AI agent, such as automatic payments, gasoline purchases and in-store payments, the number of payments worth delegating falls to about 11 a month.

The core of the analysis is that the user base must be very large for taking small fees on some transactions to generate meaningful revenue. It also said trust issues, such as users handing over credit cards and purchasing authority to an AI, are a variable.

Actual user experience suggests those barriers could fall. The outlet introduced that after using Muse and Instinct for about 2 weeks, it paid $17, or about 23,100 won, for a luggage storage service recommended by Instinct using a card saved in Stripe's payment service Link.

MoffettNathanson raised the possibility that, given the limits of transaction fees, Muse could in the long term develop into Meta's new advertising inventory. The analysis also said that if an AI agent provides sufficient utility, a subscription model could also be a way to monetise it.

The analysis said the contest in the AI shopping era is expected to hinge on how many transactions AI can handle and how those users are linked to revenue.