AgiBot proposes ‘Worken’ metric for robot activity, leaderobot.com reports

leaderobot.com reported on Nov. 3, 2026, that Peng Zhihui, AgiBot’s co-founder, president and chief technology officer, proposed “Worken,” a unit of robot efficiency, at AgiBot’s 2026 Innovation Day on Oct. 29. According to the article, the formula is “number of operating robots × autonomous working hours × output efficiency ratio.” It described tokens as measuring the amount of thinking done by AI, while Worken measures the amount of activity performed by robots.

AgiBot’s proposal reflects an approach that evaluates not only the number of robots but also autonomous working hours and output efficiency. Whether the metric’s definition and measurement methods are standardized will determine whether it can support comparisons based on work volume or pricing. The leaderobot.com article does not establish how the metric is being applied.

The proposal focuses not on the number of robots or manufacturing capacity, but on combining operational activity and work results into a single measure. The formula reported in the article multiplies the number of operating robots and autonomous working hours by the output efficiency ratio. It can therefore be read as an approach that includes “how much robots operate in the field and what results they produce” in its evaluation, which simple shipment figures can have difficulty capturing. However, this is a proposed metric; the article does not indicate that it has been adopted as a common standard or applied to actual pricing. We have covered secondary reporting on revenue and shipments in humanoid.guide’s report on AgiBot’s revenue and humanoid shipments, and reporting on the evaluation of robot work volume in Forbes’ report on the assessment of robot work volume involving AgiBot. The Worken proposal connects these discussions of business scale and work volume to the question of which variables should be used to measure them. However, the summaries of the earlier articles do not establish a direct connection with Worken or show that the evaluation methods are consistent. For the metric to allow comparisons across the industry, each element of the formula must be defined in a measurable way. The article does not say, for example, whether “number of operating robots” includes machines on standby, how “autonomous working hours” are distinguished from time involving human intervention, or what the “output efficiency ratio” is measured against. How to account for task difficulty, quality and safety-related stops would also matter when comparing figures from different sites. Unless these points are defined, measurements using the same name, Worken, may not mean the same thing, and it will be difficult to determine whether they can serve as a basis for pricing or performance evaluation. The next points to watch are whether definitions, units of calculation and verification procedures are made public, and whether the same method is applied to robots and sites beyond AgiBot.