Meta AI Chief on Muse Agent: Improving Memory and Complex Tasks, Weighing Transaction-Based Monetization

Meta's chief AI officer, Alexandr Wang, discussed the company's plans for its AI agent Muse in an interview reported by Business Insider on October 9. He said Meta will keep improving the agent's memory and ability to handle complex tasks, while reminding users that they also need to put in effort if they want the agent to make money.

Wang noted a popular type of prompt in which users tell Muse to “earn $10,000.” He said that simply making such a request does not guarantee the agent will fulfill it, and that users must contribute effort as well. He previously encouraged users to use Muse for finding ways to save money through a challenge, and noted that using the agent to earn money is a higher-level ask.

On the technical side, Wang said the memory of an AI agent differs from human memory, and that improving memory is a key focus so Muse can help users complete more complex tasks. He also pointed out that much of the industry is focused on improving coding abilities of agents, but that most people worldwide do not code and may not want to, so Meta aims to make the agent experience more accessible to ordinary users.

One month after launch, Muse remains the top free app on Apple's App Store and has become a popular product from the superintelligence lab team led by Wang. He mentioned that Meta offers tiered subscription services for Muse, but has not spent much time considering ads. He said that if users shop with Muse, or Muse helps users save or earn money, Meta may explore charging fees from those transactions.