Robots · Humanoids

Agility Digit teleoperation: who handles the exceptions on a deployed humanoid

Digit is the Western humanoid with the longest record of paid commercial work. Agility has said little about remote operation, so this page sticks to what is published and to what any deployed fleet needs around it.

Agility Digit, Bipedal warehouse and logistics humanoid
AI-generated illustration of the Agility Digit. For product photos, see Agility Robotics.
Height and weight (Digit 5)
1.81 m, 129 kg
Reach
2.2 m
Payload
Repeated lifts up to 22.7 kg (50 lb)
Battery
90 min run time, 9 min charge (10:1 run-to-charge)
End-effectors
Swappable, ISO-standard mounting flanges
Safety
Human detection, visual and audio cues, independent safety controller
Fleet software
Agility Arc (uptime, throughput, MTBI; WMS and MES links)
Operating record
Digit 4: more than 65,000 hours across customer sites

Updated

Agility's Digit is a bipedal humanoid built for one kind of work, moving totes and materials in warehouses and factories, and it has done that work for customers longer than any other humanoid. Agility says Digit 4 has logged more than 65,000 hours across sites including GXO, Schaeffler, Amazon and Toyota Motor Manufacturing Canada, and that Digit has nearly three years of continuous commercial deployment. Teams searching for "Digit teleoperation" are usually asking a deployment question, not a research one: when a production humanoid gets stuck at 3 a.m., who gets it moving again?

Can you get a Digit?

Yes, through Agility's sales team, and only as a commercial deployment. There is no store and no public price. Agility's first well-documented commercial model was Robots-as-a-Service: in 2024 GXO and Agility announced what was described as the first humanoid RaaS contract, expanding Digit's work at a GXO facility outside Atlanta, per the Agility Robotics entry on Wikipedia. Agility unveiled Digit 5 on September 15, 2026, citing more than $300 million in multi-year orders as of May 2026. Early access to Digit 5 is expected in the first half of 2027 and general availability by the end of 2027, and for the first time Agility plans deployments outside North America, starting with the EU and UK.

The company is also going public through a merger with Churchill Capital Corp XI; Agility's newsroom records the SEC declaring the S-4 registration effective on October 13, 2026.

How Digit got here

Agility spun out of Oregon State University's Dynamic Robotics Lab in 2015 and first sold Cassie, a two-legged research robot. Digit, a full-size humanoid with a torso, arms and perception, followed in 2019. Ford bought a later version in 2020 to explore last-mile delivery, and in 2021 Agility refocused Digit on warehouse and logistics material handling. The current Digit generation debuted at ProMat in 2023, shortly before Agility announced RoboFab, its humanoid factory in Salem, Oregon. Amazon began testing Digit in late 2023, and GXO put it to work at its Spanx facility near Atlanta before signing the 2024 RaaS contract. That history matters for operators: Digit has spent years on one narrow, repetitive job, which is exactly the kind of work where exceptions are predictable enough to write procedures for.

What Digit 5 changes

Reach and payload

Reach goes to 2.2 m from 5.5 ft on Digit 4, and payload rises 40 percent, to repeated lifts of up to 50 lb. That moves Digit from tote handling toward depalletizing, palletizing and machine tending, which Agility lists as target workflows.

Charging

A 90-minute battery that charges in 9 minutes, a 10:1 run-to-charge ratio against 2:1 before. On October 15, 2026 Agility named Nyobolt as its fast-charging battery partner for Digit 5.

Tools

Swappable end-effectors on ISO-standard flanges, so one robot can change grippers for different jobs.

Safety

Agility calls Digit 5 its first humanoid "engineered for cooperatively safe work at scale", with human detection, visual and audible safety cues and an independent safety controller, and says Digit was the first humanoid to pass an independent field evaluation on a customer line against OSHA-administered standards. The product page notes that specifications are preliminary and some safety features are still in development.

Is Digit teleoperated?

Agility has not publicly described a teleoperation or remote-assistance mode for Digit. Its Digit 5 materials describe autonomous workflows expanded by its "physical AI stack", fleet management through Arc, system health and diagnostics flowing to the cloud, and "a service ecosystem that was developed to maximize uptime". There is no public SDK, VR stack or developer program. We will not guess at what Agility runs internally; this page is about the layer any operator of a warehouse humanoid fleet has to plan for, whoever supplies it.

Arc and the exception problem

Arc is Agility's cloud platform for deploying and managing Digit fleets. It reports uptime, throughput and Mean Time Between Incidents, and connects to existing warehouse and manufacturing execution systems. The last metric is the one that shapes staffing. Every incident is a moment when the robot cannot continue on its own: a tote is crushed or mislabeled, a conveyor is jammed, a pallet is stacked off pattern, a person is standing in the robot's path. A fleet that runs 20 hours a day across several sites will have incidents on every shift, and someone has to clear each one.

On-site associates

Clear physical problems the robot cannot: a fallen load, a damaged tote, a blocked aisle.

Remote supervisors

Watch fleet dashboards and live views, acknowledge alerts, and decide whether to retry, reroute, skip the item or call someone on site.

Remote operators

Where a robot exposes a remote control or assist interface, finish the task through it. On a closed platform that is only possible through the maker's tools.

Engineers

Review the incident log, so recurring failures become software fixes or training data rather than permanent work for people.

The split between those roles is a design decision for each site, and it changes as the robots improve: early in a deployment, more incidents need a person; later, the remote desk handles more and the site handles less. Our human-in-the-loop guide for humanoid robots describes how teams structure that exception desk and turn its sessions into correction data.

What an exception desk does on a tote-handling fleet

Answers every alert within an agreed time, because a stalled robot in a tote loop can back up a conveyor in minutes.
Follows a written decision tree per incident type: retry, skip and flag, hand to site, or stop the robot.
Logs each incident with the robot, location, task and resolution, so MTBI and its causes are visible to engineering.
Respects the safety system. If the robot has stopped for a person, the remote role is to confirm, not override.
Keeps the same people on the same site, so they learn its quirks: which station mislabels, which aisle gets crowded at shift change.

Why Digit 5 raises the staffing question

The 10:1 run-to-charge ratio is the change that matters most for the people around the robot. At 2:1 a robot spends a third of its day charging, and the gaps give a site natural slack. At 10:1 charging drops to under a tenth of each cycle, and Agility's product page lists 20 working hours in every 24. More hours of work means more incidents per robot per day, spread over every shift, including nights and weekends when a site has the fewest people. Wider workflows add to it: depalletizing and machine tending have more ways to go wrong than moving empty totes. The result is that exception coverage, not robot count, often becomes the limit on how many hours a fleet really runs.

Questions to ask any warehouse humanoid vendor

Which incidents does the robot recover from on its own, and which need a person?
Who is that person by default: your associates, the vendor's remote team, or a third party you choose?
What can a remote person do through the vendor's tools: acknowledge, retry, reroute, or drive the robot?
What response time does the service level assume, and on which hours?
How are incidents logged, and do you get the data, including the video and robot state around each one?
Can a third-party operator team work inside the vendor's tools under your contract?

The answers decide how much of the exception desk you staff yourself. Our guide to robot teleoperation as a service covers how outsourced operator teams are usually contracted.

Latency for supervision versus control

Exception handling is mostly supervisory: read the state, choose an action, confirm. That tolerates far more network delay than direct teleoperation, so a remote desk can sit in another region or time zone and still be effective, as long as the robot's own controllers handle balance and motion. Direct remote control of a full-size biped carrying 50 lb is a different matter, and needs a short round trip plus local safety limits. Our teleoperation latency guide covers where each kind of task falls.

Digit compared with humanoids built for data collection

Digit and research humanoids answer different needs. Digit is a closed industrial product sold with fleet software and service, aimed at a defined set of logistics workflows. You do not program it yourself or record your own teleop datasets on it. If your goal is to train your own whole-body or manipulation policies, you need an open platform with an SDK and a teleop stack, such as the Unitree G1, on which Unitree's open xr_teleoperate records episodes directly, or a full-size Unitree H2 for adult-height reach. If your goal is to run humanoids in a warehouse this year with a vendor behind them, Digit is one of very few options with a commercial record. Our warehouse and factory page covers the wider automation picture.

How RoboPilots can support a humanoid fleet

We do not have a partnership with Agility, and nothing here implies one. What we offer any humanoid program is the people layer. Several humanoid makers have sent RoboPilots robots to test our operating protocols. For a customer fleet we train a dedicated group of operators on your own units, your sites and your incident protocol, and keep the same group on your program.

One-week trial

Sets the baseline on your robots, whether that is clean episodes per operator hour for data collection or response and resolution times for exception handling.

Coverage

Shifts across time zones, including US Pacific, so a 20-hour operating day has a person on every hour of it.

Data

Interventions and teleop sessions logged by task and outcome and delivered as LeRobot, RLDS or HDF5 where the platform allows recording.

Interfaces, specs and sources

  • Agility Arc fleet cloud
  • Warehouse and MES integration
  • Swappable ISO-flange end-effectors
  • No public SDK or teleop stack

Availability: Sales-led, no public price. Digit 4 is deployed commercially at customer sites in North America (Agility names GXO, Schaeffler, Amazon and Toyota Motor Manufacturing Canada). Digit 5 was unveiled on September 15, 2026, with early access expected in the first half of 2027 and general availability by the end of 2027, adding the EU and UK to the US and Canada. GXO signed a Robots-as-a-Service contract in 2024.

Specs are the maker's published figures, checked on . RoboPilots is not affiliated with Agility Robotics; product names are trademarks of their owners.

Frequently asked questions

How much does Agility Digit cost?
Agility does not publish a price. Digit is sold through Agility's sales team as a commercial deployment, and GXO signed what was described as the first humanoid Robots-as-a-Service contract with Agility in 2024.
Is Agility Digit teleoperated?
Agility has not publicly described a teleoperation or remote-assistance mode for Digit. Its materials describe autonomous workflows, fleet monitoring through Agility Arc, and a service ecosystem for uptime.
Where is Digit deployed?
Agility says Digit 4 has logged more than 65,000 hours at customer sites in North America, including GXO, Schaeffler, Amazon and Toyota Motor Manufacturing Canada.
When is Digit 5 available?
Agility expects early access to Digit 5 in the first half of 2027 and general availability by the end of 2027, with the EU and UK added to the United States and Canada.
What is Agility Arc?
Agility's cloud platform for deploying and managing Digit fleets. It reports uptime, throughput and Mean Time Between Incidents and connects to warehouse and manufacturing execution systems.