Boots on the Ground: A Field Guide to Humanoid Robot Use Cases and Live Deployments, 2025–26

Overview Humanoid robots moved out of labs into demonstrable pilots across logistics, manufacturing, healthcare and retail in 2024–25. Deployments remain limited and task-specific, but the traction is real: companies are trialing bipedal platforms for picking, line‑tending and patient assist where a human form factor provides an advantage over traditional mobile robots and cobots. This survey synthesizes public announcements, company materials and on‑the‑record customer feedback to catalogue where humanoids are actually being used today and what organizations should expect when evaluating pilots. Where humanoids are being trialed Warehouses & logistics Use case: package handling, shelf replenishment, pallet handling and piece‑picking where doorways, shelving heights and human‑designed workstations favor a bipedal form factor. Agility Robotics — Digit: Agility positions Digit for fulfillment centers and last‑mile staging. Public pilots focus on box handling and handoffs to existing conveyors and sortation systems. Agility emphasizes the robot’s human‑scale reach and stair‑capability as reasons warehouses test Digit where AMRs and stationary automation cannot operate easily. 1X Technologies — NEO / EVE: Marketed explicitly as warehouse assistants, 1X runs pilot programs that integrate with picker workflows (assistive handoffs, simple shelf retrieval). Company materials highlight slot‑level tasks and overnight restocking pilots with logistics integrators. Manufacturing Use case: line tending, small‑parts assembly, simple fastening and inspection tasks that require human‑like manipulation and access to existing tooling. Figure AI: Figure is positioning humanoids as production‑first robots — demonstrations concentrate on repetitive production operations that benefit from human kinematics. Their pitch: faster line integration than custom tooling because the humanoid can use existing human jigs and stations. Apptronik: Focused on industrial integrations, Apptronik markets torque‑capable limbs for lifting, handling and machine tending. Public demos show the system operating end‑to‑end with human colleagues in pilot cells. Tesla (Optimus): Tesla has publicly framed Optimus as both a factory worker and an internal R&D platform. Tesla’s stated in‑house trials emphasize logistical tasks and repetitive handoffs inside assembly areas, designed to validate internal workflows before external sales. Healthcare & eldercare Use case: mobility assistance, patient transfer aid, monitoring and companionship trials in long‑term care. Safety and regulatory scrutiny are highest here. Sanctuary AI: Marketed as a general‑purpose humanoid for assisted tasks, Sanctuary has spoken publicly about healthcare trials focused on non‑clinical assistance (transport of supplies, fetch tasks) while deferring clinical patient handling to specialized devices. UBTECH: Centred on social and eldercare robots in Asia, UBTECH runs trials in senior care facilities for monitoring, reminders and companionship features while limiting physically risky transfers. Xiaomi — CyberOne: Presented as a consumer/legal assistant prototype and marketed to highlight human‑centric interaction; Xiaomi’s public demos remain at the prototype stage, focusing on sensing and social features rather than clinical tasks. Commercial & retail Use case: shelf stocking, store concierge and in‑store logistics where a human shape eases customer interaction or navigation in constrained retail aisles. Early trials use humanoids for night stocking and concierge duties; companies emphasize branding and customer experience as primary ROI for in‑store pilots. What companies are promising — and how they position robots Two positioning narratives dominate: Task replacement in constrained, human‑designed spaces: Agility, 1X and Figure emphasize humanoids can be dropped into human stations or traverse stairs and doorways where wheeled robots fail. Human‑machine partnership and augmentation: Apptronik, Sanctuary and UBTECH pitch humanoids as teammates that reduce physical strain and extend human teams rather than fully replace them, particularly in care settings where empathy and safety matter. Pilot outcomes, customer feedback and pain points Publicly available performance data is limited; what exists is primarily company‑reported metrics and early customer anecdotes. Common themes: Uptime and reliability: Early pilots report lower continuous uptime than mature AMRs — battery swaps, sensor recalibration and software hiccups typically limit shifts to partial day use. Companies describe steady improvements between iterations. Cycle time and throughput: For constrained, repetitive tasks (box transfer, single‑screw fastening), humanoids can achieve cycle times competitive with new human hires after task scripting. However, variability in unstructured environments increases cycle time and error rates. Handoffs and human‑robot workflows: Integrators stress that the costliest work is designing smooth handoffs and safety interlocks. Customer feedback repeatedly cites integration engineering as the dominant part of pilot budgets. Safety and regulatory hurdles: Warehouses prioritize OSHA‑style workplace safety certification; healthcare trials face additional scrutiny around patient handling, liability and local medical device rules. Many pilots intentionally avoid full patient transfers to limit regulatory exposure. “The machine can do a slice of what our people do, but the systems work is where time gets spent.” — paraphrased feedback from a logistics integration lead involved in multiple humanoid pilots (public commentary, anonymized). Cost / business models Companies use a mix of pricing approaches: outright sale, lease and Robotics‑as‑a‑Service (RaaS) with per‑hour or per‑task fees. Directional cost/benefit observations: Unit economics are still early: up‑front hardware/software and integration costs commonly exceed six figures in pilot scopes when integration is included. Value cases tend toward augmentation: reducing overtime, filling risky night shifts and reducing workplace injuries — not immediate full headcount replacement. Competitive comparison (high level) Company Primary focus Strength Typical pilot task Agility Robotics (Digit) Logistics Human‑scale mobility, stair access Box handling, conveyor handoff 1X Technologies (NEO/EVE) Warehouse assistance Integration with picker workflows Restocking, shelf retrieval Figure AI Production Line‑integration focus Small‑parts assembly, machine tending Apptronik Industrial integrations Torque‑capable limbs Lifting, machine tending Sanctuary AI / UBTECH Care & social robotics Interaction and safety Monitoring, assistive tasks Tesla (Optimus) Internal factory use → external Vertical integration Logistics, repetitive tasks Xiaomi (CyberOne) Consumer/assistant Sensors and social interaction Prototype assistant tasks Lessons learned for scaling Integration beats hardware: pilots show that integration engineering (fixtures, software, safety interlocks) dominates cost and schedule. Start with constrained tasks: shelf‑level pick, conveyor handoffs and single‑tool assembly tasks scale fastest. Human factors matter: staff acceptance, workspace redesign and clear safety boundaries accelerate adoption. Regulatory engagement early: healthcare pilots must budget for additional approvals and liability planning. Procurement checklist for evaluating humanoid pilots Define a constrained, measurable task (cycle time, error rate, uptime) and baseline human performance. Require vendor integration estimates: software, fixtures, safety interlocks and IT connectivity. Ask for mean time between failures, planned maintenance schedule and support SLAs. Validate safety case and insurance coverage; in healthcare, require a clinical risk assessment. Clarify pricing model: capital purchase vs RaaS vs pilot‑to‑production transition terms. Plan for staff retraining and a human‑in‑the‑loop escalation protocol. Conclusion Humanoids in 2025 are no longer science‑fair curiosities — they are focused pilots and limited deployments where the human form factor solves a geometry, interaction or access problem. Expect incremental commercialization: companies and customers will expand usage by locking down narrowly scoped, measurable tasks, investing in systems integration, and carefully navigating safety and regulatory hurdles. For procurement teams, the pragmatic approach is to pilot early, measure narrowly and budget heavily for integration rather than expecting an out‑of‑the‑box solution.