system
The only layer customers are already paying for: it solves one narrow, provable business problem, not a promise of general-purpose robots.
客户唯一已在付费的一层：垂直工作流 ROI。
Bottlenecks瓶颈 P18 Vertical workflow ROI垂直工作流 ROI
What it isWork is the vertical-workflow layer: robots or automation solving one narrow, measurable business problem for a paying customer, priced on outcome rather than on the robot itself. P18, Vertical workflow ROI is the taxonomy's home for that thesis: that customers buy outcomes, not a general-purpose robot that can do any job (what the industry calls 'physical AGI'). Work is kept separate from Integration (P17, Integration, uptime, service, its own system below) because a workflow can pencil out on a spreadsheet while the underlying deployment is still unreliable in the field: two different failure modes with two different owners.
是什么“工作”（Work）是垂直工作流层：机器人或自动化为付费客户解决一个狭窄、可度量的具体业务问题，按结果而非按机器人本身定价。P18，垂直工作流 ROI 是这一论点在分类法中的落脚点——客户购买的是结果，而非 physical AGI。“工作”与“集成”（P17，集成、在线率与服务，下文另立一个系统）被分为两个独立系统，是因为一条工作流可以在表格上算得通，而其底层部署在现场仍不可靠——这是两种不同的失败模式，也对应两个不同的责任方。
ThesisP18, Vertical workflow ROI ranks 1st of 18 problems on CE-N (the strongest capital-efficiency position on the board), driven by the highest budget score of any problem tracked (5.0/5, meaning more money is already authorized to be spent on this than on anything else here). This is where the thesis lives most strongly, because it is the one place customers are already paying for outcomes rather than funding a robotics science project. That budget score is real in a handful of narrow verticals (agricultural weeding, automated food service, constrained warehouse tasks) where payback is documented, but it is not yet real in general-purpose humanoid industrial deployment, where task success rates still run below 50% and industrial use remains a single-digit share of shipments. Most of the actual exposure to it is private, not public, and a portion of the private-book budget score is circular commitment between related balance sheets rather than external recurring revenue. Size conviction to the instrumented, narrow-vertical cases, not the headline TAM (broad total-addressable-market estimates, which tend to run far larger than what's actually being spent today).
论点P18，垂直工作流 ROI 在 CE-N 下位列 18 个问题中的第 1 名，驱动因素是全站最高的预算得分（5 分制 5.0 分）——这正是本论点最坚实之处，因为这是客户已经在为结果买单、而非在资助一个机器人科研项目的唯一环节。这一预算得分在少数几个细分领域是真实的——农业除草、自动化餐饮、受限场景下的仓储任务——客户已在付费，回本周期有据可查；但在通用人形机器人的工业化部署中尚未成立：高难度任务的成功率仍低于 50%，工业用途在出货量中仍只占个位数百分比。对它的实际敞口大多是私募，而非上市公司，且私募账本中的部分预算得分是相关资产负债表之间的循环承诺，而非外部经常性收入。应把信心押注在已被量化验证的细分垂直案例上，而不是标题式的 TAM 数字。
Products mapped to this system映射到本系统的产品
Robots, AMRs, and fleet-management software that automate warehouse fulfillment and logistics material movement, several of them priced against measured throughput or pick-productivity outcomes rather than sold as generic hardware.
5 companies 5 家公司 +1 named in conference materials, unverified +1 项来自会议材料，未经核实
Agricultural automation named in the conference doc as end-market automation for farm operations.
+2 named in conference materials, unverified +2 项来自会议材料，未经核实
Surgical robotic systems sold on a vertical-workflow basis rather than as generic robot hardware, spanning an on-site platform/instruments/service business plus additional surgical robotics named in the conference doc.
1 companies 1 家公司 +1 named in conference materials, unverified +1 项来自会议材料，未经核实
Autonomous vehicles and mobility devices aimed at passenger and freight transportation, spanning an on-site autonomous aerial vehicle operator plus additional autonomous mobility, trucking, and wearable-mobility companies named in the conference doc.
1 companies 1 家公司 +3 named in conference materials, unverified +3 项来自会议材料，未经核实
Ground and aerial robots for industrial inspection, hazardous-environment response, and defense/security applications, spanning on-site quadruped/humanoid and uncrewed-aircraft platforms plus additional inspection and aerial-application companies named in the conference doc.
2 companies 2 家公司 +5 named in conference materials, unverified +5 项来自会议材料，未经核实
Robotic automation of specific manufacturing and construction processes -- welding, process automation, and construction robotics -- named in the conference doc.
+3 named in conference materials, unverified +3 项来自会议材料，未经核实
Companies mapped to this system映射到本系统的公司
The live-web research behind this system concentrated on the neighboring Integration system (Teradyne, Agility Robotics, Symbotic, and Intuitive Surgical; see Integration, below), because that is where this site’s named public and private proxies actually sit.
Direct primary evidence on P18, Vertical workflow ROI’s own vertical-workflow economics is thin. Almost everything below on P18, Vertical workflow ROI itself comes from the sell-side pass, not from primary sources.
Sizing estimates for the opportunity span nearly two orders of magnitude:
Deployment reality checks against those numbers:
Where a real ROI loop exists today, sell-side reporting shows it is narrow and rarely humanoid:
These are the cleanest evidence in the batch that vertical workflow ROI is real, and none of them is a general-purpose humanoid. Two more figures get cited as if they were revenue, but aren’t: Symbotic’s $432bn warehouse SAM and AutoStore’s ~$600bn AS/RS TAM are both company-disclosed addressable markets from investor days, on the same footing as Agility Robotics’ ”>$300m of contracted orders”: committed capital and addressable market, not received revenue.
P18, Vertical workflow ROI’s rank-1 score is real in a handful of narrow verticals (agricultural weeding, automated food service, constrained warehouse tasks) where the customer is already paying and the payback is documented. It is not yet real in general-purpose humanoid industrial deployment: success rates sit below 50% on the hard tasks, and industrial use is still a single-digit share of shipments.
There’s a second risk underneath the 5.0 budget score itself. Research on the Integration system, below, finds that several of the largest named private-book “customers” (Schaeffler, Foxconn, CATL) are also investors in, suppliers to, or hosts for the company they’re buying from. Some of that budget score is circular commitment between the same balance sheets, not external recurring revenue.
Bottom line: size conviction to the instrumented, narrow-vertical cases. Treat the humanoid-industrial TAM figures as a forecast, not a result.
支撑本系统研究简报的一手网络调研，主要集中在相邻的“集成”系统上——泰瑞达、Agility Robotics、Symbotic 与直觉外科（见下文“集成”部分）——因为本站具名的上市与私募代理公司主要就落在那里。
专门针对 P18，垂直工作流 ROI 自身垂直工作流经济性的一手证据相当有限。下文关于 P18，垂直工作流 ROI 的内容几乎全部来自卖方研究批次，而非一手来源。
卖方研究对这一机会的估算跨度接近两个数量级：
对照这些数字的部署现状核查：
在当下真实存在投资回报闭环的地方，卖方报告显示这些闭环都很狭窄，且很少是人形机器人：
这些是整个样本中关于垂直工作流 ROI 真实存在的最清晰证据，而它们无一是通用人形机器人。另有两个数字被当作收入引用，实则并非收入：Symbotic 的 4320 亿美元仓储 SAM 与 AutoStore 约 6000 亿美元的 AS/RS TAM，均为公司在投资者日自行披露的可触达市场数字，与 Agility Robotics“超过 3 亿美元的已签订单”属于同一性质——是承诺资本与可触达市场，而非已实现收入。
P18，垂直工作流 ROI 的第一名评分，在少数几个狭窄垂直领域是实至名归的——农业除草、自动化餐饮、受限的仓储任务——在这些场景中，客户已经在付费，回收期也有据可查。但在通用人形机器人的工业部署上，这一评分尚未被真正兑现：高难度任务的成功率低于 50%，工业用途仍仅占出货量的个位数百分比。
预算得分本身之下还潜藏着第二重风险。下文关于“集成”系统的研究发现，私募名录中最大的几个具名“客户”——Schaeffler、富士康、宁德时代——同时也是它们所购买公司的投资者、供应商或部署场地。部分预算得分实为同一批资产负债表之间的循环承诺，而非外部经常性收入。
结论：信念的强弱应锚定于那些有据可查、狭窄垂直的案例。把人形机器人工业化的 TAM 数字当作预测，而非既成结果。