Manus

For ToTheMoon, Manus is a test of what separates an agent from a chatbot's polished answer. The channel first examined it in March 2025 as a Chinese system that takes a task, gathers data, works through several steps and hands back a finished result; the hosts stressed that the packaging is what creates the sense of an agent, whereas ChatGPT demos end in a nicely written reply (episode 049). Since then the question has been whether Manus sells model intelligence or completed work, and what its hype is worth.

The first review was enthusiastic: access was invite-only, yet the system was described as a leap that combines deep research, generation, computer operations and long memory, with output coherent enough to be compared to an MBA-level consultant (episode 049). A week later the tone turned to verification: the hosts asked whether Manus was real (episode 050) and passed along the rumor that Claude from Anthropic might be running under the hood (episode 051). By April the doubt had settled: Manus looks impressive from the outside, but it is unclear where it is heading or why it drew more attention than deep research (episode 054).

In 2026 the story became Meta's acquisition. The hosts read the deal as a distribution play: a strong model without access to products and data stays a laboratory, while Meta can embed agency in the everyday actions of billions of people (episode 091). The channel then noted the first post-deal product, Manus Meetings (episode 095), and in summer recalled the whole arc, including China barring the company's key people from leaving (episode 136).