To address this, ZhiYuan has combined Work+Token to introduce a new measurement unit called Worken (Effective Intelligent Working Hours). Previously, a robot company's shipment volume, profit, and market share could serve as standards for external evaluation. Under various metrics, each company has the opportunity to claim a leading position. Shipment volume assesses the production capacity of the robot company, profit evaluates its business model, and market share measures the actual sales volume. However, no single metric can measure how much work a robot has actually completed, whether it can execute commands correctly, or whether it can work continuously. These factors require a unified evaluation standard.
ZhiYuan Introduces New Robot Evaluation Standard Worken
For example, if a robot is tasked with screwing in a screw, it must align the screwdriver with the screw slot and turn the screw into the target position at the correct angle and number of turns. During this process, issues may arise such as the screwdriver not being aligned with the slot, the screw being misaligned, or the need for on-site reasoning calculations in different scenarios, leading to excessive time consumption. These could all result in command execution failures or instability in operation. ZhiYuan co-founder, co-president, and head of the embodied business system Yao Maoqing pointed out that embodied intelligence is entering a more realistic test: the standard for evaluating robots is no longer whether they can perform a demonstration but whether they can operate stably in real environments 24/7 and respond to disruptions.Moreover, it is necessary to calculate ROI (Return on Investment). This means that the evaluation of robots must take place within actual work scenarios. Only by comprehensively considering the quantity of actual use, autonomous working time, and unit workload can we get closer to the true capabilities of robots.
The Worken Calculation Formula Will Change the Evaluation Method in the Robotics Industry
Based on this idea, ZhiYuan proposed the Worken calculation formula: Worken = Number of Active Robots × Autonomous Working Hours × Output Efficiency Ratio. This calculation method aims to further assess whether robots can autonomously and continuously complete valuable work, beyond just the number of robots manufactured, and how much actual output they generate. In simple terms, if there are 10 robots, each working autonomously for 6 hours, with a qualified output per unit time being half of the human benchmark, then the Worken for these 10 robots would be 10 × 6 × 0.5 = 30.According to this setup, the amount of tasks they completed is equivalent to 30 human benchmark hours. It is important to note that this is a measure of task volume, not the actual number of employees replaced. This way, the working capability of robots will be more intuitively reflected, and buyers will have a clearer method for calculating the input-output ratio of robots, making it quantifiable how much was spent, what results were produced, and what costs were saved.
The first two data points in the formula can be derived from simple calculations, but the output efficiency ratio is more challenging to determine. ZhiYuan partner, co-president, head of the central R&D department, and head of the interactive business system Wang Chuang admitted in a media interview that the last indicator of Worken, the output efficiency ratio, remains relatively vague at this stage.This vagueness primarily exists in tasks that are difficult to quantify. For example, in the loading and unloading process in factories, metrics like cycle time and pass rate can be compared, but in hotel reception, it is challenging to measure solely by the number of interactions: a robot may be able to answer questions all day but cannot take on other responsibilities of front desk staff. How different tasks are converted remains to be standardized. Therefore, until the output efficiency ratio is further defined and calculated, Worken remains a conceptual framework that cannot yet participate in actual calculations. What we need to see is that its introduction challenges the valuation benchmarks of most robot companies and proposes a more easily computable delivery concept and comparison standard.
Currently, robots have moved from laboratories into the public eye, preparing to enter various job positions, with user attention shifting from demonstrations to real scenarios. In this trend, both companies purchasing robots and ordinary consumers who will use robots in the future need a metric that can intuitively reflect the capabilities, performance, and user experience of robots. While Worken may not be widely adopted in the industry in the short term, similar definitions are indeed what the robotics industry currently needs.
