On Nov 11, Moonshot AI released hosted agents. An analysis published by TMTPost that day noted that the market mostly read this as a "feature upgrade," but argued it is not a product iteration but a restructuring of employment forms — AI vendors are moving from "selling tokens" to "renting digital employees."
Hosted agent launch: from feature upgrade to employment restructuring
According to TMTPost, the hosted agents released by Moonshot AI were widely seen by the market as a feature upgrade. The article takes a different view, arguing that the essence is a restructuring of employment forms. Based on the principle of "dynamism," it proposes a three-tier rent-or-buy decision framework and reveals that the pricing anchor is sliding from IT budgets to labor budgets.
This article's judgment: this is not a product iteration but a restructuring of employment forms — AI vendors are moving from "selling tokens" to "renting digital employees."
The analysis further notes that in general-purpose scenarios, "renting" is the main channel, but "buying" will not disappear; vertical scenarios enter a hybrid state. That means enterprises will acquire AI capabilities in more diverse ways rather than simply shifting to rental.
Same-day salon echoes: from "selling tokens" to "selling results"
On the same day, QbitAI reported on the "Tech with Lenovo · Token Fantasia" offline salon held in Beijing on Oct 30. Themed "from selling tokens to selling results," the salon debated three major issues of the token economy. Although the report did not disclose the specifics of the discussion, its theme echoes TMTPost's analysis of Kimi's hosted agents — both point to AI vendors' business models shifting from token-based billing to billing by results or "digital employees."
Pricing anchor migration: from IT budgets to labor budgets
The core judgment of the TMTPost article is that the Agent pricing anchor is sliding from software budgets to labor budgets. This shift means AI vendors' charging logic is no longer confined to IT procurement but is beginning to benchmark against human resource costs. If the trend holds, the path of AI commercialization will change significantly, and corporate budget allocation may adjust accordingly.
The analysis also stresses that rent-or-buy decisions must rest on the principle of "dynamism," with form choices differing by scenario. General-purpose scenarios are mainly rental, vertical scenarios show a hybrid state, and outright purchase will still exist.
What to watch next
Moonshot AI has not yet provided further details on the specific pricing and business model of its hosted agents. The market is watching how the hosted agents will be priced, how receptive enterprises will be, and whether the "renting digital employees" model can work in both general-purpose and vertical scenarios. Both TMTPost and QbitAI reports show that Agent commercialization is evolving from the token economy to the results economy, and how this trend plays out on the ground warrants continued observation.