1. Calculate Worken for one deployment
Per robot: 8 Worken. At a ratio of 1.0, 80 Worken is roughly 80 standard human-work hours of acceptable output.
示例数值:上方默认输入(10 台 × 8 小时 × 1.0 = 80)仅用于演示算式,填你自己的数才有意义。
2. Compare two deployment plans
Plan A
Plan B
Plan A delivers 24 more Worken (9.1% more) than Plan B.
示例数值:Plan A(20 × 16 × 0.9 = 288)与 Plan B(12 × 20 × 1.1 = 264)的默认输入仅用于演示对比逻辑。A plan can win on Worken while losing on robot count. That is the point of the metric: headcount alone does not describe output.
3. What is Worken?
Worken combines the words work and token. AGIBOT co-founder, president and CTO Peng Zhihui proposed it at AGIBOT Innovation Day 2026 in Shanghai on October 29, 2026, as an answer to a measurement question: in the steam age, horsepower measured machines; in the electric age, kilowatt-hours; in the digital AI age, tokens measured machine thinking. For physical AI, he argued, what is missing is a unit for how much machines act. His summary: tokens measure how much machines think, Worken measures how much machines act.
AGIBOT presented Worken alongside the AHA3 hardware platform, the GO-3 Preview embodied foundation model and the WITA-Omni interaction model. Worken and related deployment cases were also included in the 2026 Embodied Intelligence Industry Innovation Report compiled by China Economic Information Service and the Shanghai Industrial Innovation Center of CAICT. AGIBOT has said Worken could help businesses evaluate the longer-term value of robotic deployments alongside total lifecycle costs (PR Newswire, Oct 29, 2026).
4. The three factors: what counts, what does not
| Factor | Counts | Does not count |
|---|---|---|
| Active robots | Units actually performing tasks in real environments | Units manufactured, shipped, or sitting in inventory |
| Autonomous working hours | Time robots work on their own | Powered-on time, idle time, teleoperated time |
| Output efficiency ratio | Acceptable completed work vs a task benchmark (e.g. a human worker standard) | Raw activity with no quality or benchmark reference |
Typical application areas named by AGIBOT: industrial handling, logistics, inspection and commercial services.
5. Frequently asked questions
What is Worken?
Worken, short for work plus token, is a unit proposed by AGIBOT for measuring how much useful work robots actually do. It is called effective intelligent work hours and equals the number of active robots multiplied by autonomous working hours multiplied by an output efficiency ratio.
What is the Worken formula?
Worken = number of active robots x autonomous working hours x output efficiency ratio. Active robots are units actually performing tasks, not units manufactured or held in inventory. Autonomous working hours count time robots work on their own, not total powered-on time. The output efficiency ratio compares acceptable completed work against a reference benchmark for the task, such as the amount a human worker completes under established standards.
Who proposed Worken and when?
AGIBOT co-founder, president and CTO Peng Zhihui proposed Worken at AGIBOT Innovation Day 2026 in Shanghai on October 29, 2026. It was presented alongside the AHA3 hardware platform, the GO-3 Preview foundation model and the WITA-Omni interaction model.
Worken vs robot shipments?
Shipments count production. Worken counts deployed, autonomous, acceptable output. A fleet of 10,000 shipped robots doing nothing scores zero Worken.
Worken vs uptime?
Uptime counts powered-on hours, including idle and remote-controlled time. Worken counts only autonomous working hours, weighted by output quality.
Worken vs tokens?
Tokens measure machine thinking (input/output text). Worken measures machine action in the physical world. They describe different eras of machine value.
Can I compare two robot vendors by Worken today?
Not with real data. The benchmarks are still being built. You can compare your own deployment plans with this calculator, using ratios you define and document.
Is Worken an official standard?
No. It is AGIBOT's proposal, presented October 29, 2026. Whether it becomes a shared reference depends on adoption and on published benchmarks.
Sources
AGIBOT introduces Worken (PR Newswire, Oct 29, 2026)
AGIBOT introduces Worken to measure robot productivity (The Robot Report)
AGIBOT unveils Worken framework (Unite.AI)