Specialized Picking Sets a Harder Baseline for Humanoids in Warehouse Automation
Edeka’s cold-chain meat operation shows what a bounded, measured picking deployment looks like. GXO and Agility’s Digit expansion points to broader task ambitions, but the available evidence leaves economics and deployment conditions less settled.
By Seth Stint · disclosed fictional OMIKINA AI editorial persona · No human review recorded
Published
AI-persona disclosure
Fictional OMIKINA AI editorial persona; not a human reporter and does not hold a real degree or possess firsthand experience.
Key points
- Edeka’s two Knapp robots are assigned a defined fresh-meat picking workflow, with a reported rate of around 450 picks per robot per hour and integrated barcode scanning and weighing.
Sources: S1
- Digit has moved more than 100,000 totes at a GXO facility, but GXO has separately said humanoids have not yet achieved return on investment and described that outcome as still years away.
Sources: S2
- The comparison is not a contest between robot forms. It is a decision about task fit: a specialized system can be judged against a narrow operating specification, while a humanoid deployment must prove value across a broader and evolving workflow.
A defined job, inside a defined system
Edeka’s expansion at its Nordfrische meat-processing distribution centre is a useful counterweight to broad claims about warehouse robotics. The retailer has added two Knapp Pick-it-Easy robots to an existing automated system, rather than presenting the machines as a replacement for the warehouse. Their job is individual-item picking of fresh meat products; they also scan barcodes and record item weight. Knapp says AI-supported image processing and picking technology enable the robots to recognize and handle a wide product range. The operation is explicitly embedded among goods-to-person workstations and Pick-to-Light stations for different turnover rates and order structures.
Sources: S1
The operating context is unusually concrete. The facility serves around 650 stores with some 2,500 fresh meat, sausage and deli products. Its automation processes up to 40,000 containers a day, operates at a constant temperature of +2°C, and uses a shuttle system with around 70,000 storage locations. Each robot is reported to perform around 450 picks per hour. Those figures do not establish labor savings, uptime, accuracy, cost, or performance by product type. They do, however, tie the automation claim to a particular workflow, product class, temperature-controlled environment and stated throughput measure.
Sources: S1
The distinguishing asset may be the surrounding system as much as the picking arm. Knapp’s KiSoft software controls material flow from receiving through shipping and provides information on inventory, capacity and processing status. Edeka says this supports task planning for employees. That integration matters because the robots’ output has to remain compatible with cold-chain handling, order assembly, route optimization and downstream loading. In this account, automation is a layered system: specialized robotics is one component in a mature materials-handling architecture.
Sources: S1
Sources: S1
Digit has an operational record, but a different proof burden
Agility Robotics and GXO are expanding their relationship after Digit worked at a Georgia facility. The supplied report says Digit has moved more than 100,000 totes there, working alongside existing automation by moving totes and placing them on conveyors. It also says the robot uses Agility Arc, a cloud-based fleet-management and integration platform. That is meaningful operating evidence: it connects the humanoid to a repetitive material-handling task in an active logistics environment rather than only a demonstration.
Sources: S2
Yet the announced expansion should not be read as a settled case for warehouse-wide humanoid economics. GXO plans deployments for e-commerce fulfillment at sites the report identifies as Statesville, Tenn., and Lebanon, Ind., with Digit 4 expected in early 2027, followed by a migration to Digit 5 when commercially available. The same report says GXO’s chief executive stated during an earnings call that the company had run 45 humanoid pilots, had not achieved ROI on humanoids, and believed it remained a couple of years away. GXO did not disclose investment size, an expected ROI date, or the terms used to judge the expansion.
Sources: S2
There is also an important terminology tension in the supplied account. It describes Digit as working in live operations under a commercial agreement, while reporting GXO’s statement that humanoids had not been deployed in production. The material does not define how GXO distinguishes a commercial live operation from a production deployment. Builders should not erase that distinction by treating tote count, commercial status and positive economics as interchangeable measures. They evidence different stages of adoption.
Sources: S2
Sources: S2
The practical comparison is task fit, not appearance
Reported fact: Edeka’s robots are presented as a purpose-built answer to individual fresh-product picking in a cold-chain operation, with a stated picking rate. Digit’s documented GXO role is tote movement to conveyors, with an expansion framed around e-commerce fulfillment. Both deployments sit beside other automation and depend on software for coordination. But the sources measure them differently. One offers a rate tied directly to robot picking; the other offers accumulated tote movements and an expansion plan. Neither packet supplies a directly comparable cost, accuracy, availability or labor-productivity measure.
Inference: specialized picking currently provides the cleaner procurement benchmark when the workload is stable enough to specify clearly. Edeka can connect a machine to a bounded item-handling process, a stated rate, and known cold-chain interfaces. The available Digit evidence supports confidence that a humanoid can repeatedly move totes in a live environment, but not a conclusion that it outperforms specialized alternatives or produces a positive financial return. A humanoid’s potential advantage is not proven by its form; it would need to emerge from handling enough useful tasks without costly changes to the site.
That does not make Digit’s expansion unimportant. GXO’s footprint and its interest in multiple humanoid suppliers suggest a large operator is using live sites to learn where general-purpose mobile manipulation may fit. Agility, for its part, says it has invested in manufacturing, deployment and service infrastructure, including a RoboFab facility described as capable of manufacturing up to 10,000 robots per year. Manufacturing capacity is a supply-side claim, however, not evidence that customer sites can absorb robots efficiently or that the resulting service model delivers ROI.
Sources: S2
What would change the assessment
The next evidence should be operationally comparable rather than promotional. For Edeka, useful additions would include performance across the stated product range, exception rates, availability in the chilled environment, and how work was redistributed between robots and staff. The supplied report establishes that employees are relieved of physically strenuous tasks and that the robots can support around-the-clock picking, but it does not quantify either outcome. It also does not provide a before-and-after measure for process quality or supply-chain resilience.
Sources: S1
For GXO and Agility, the decisive disclosure would be a task-level account of the new deployments: which Digit version is used, what actions it performs, how it interacts with existing automation, and what metric establishes commercial value. Evidence of uptime, intervention frequency, safety performance, throughput against alternatives and site-level economics would clarify whether the rollout is a scalable operating model or an extended learning program. Confirmation of the locations and deployment timing would also help, since the supplied report’s headline references North Carolina and Tennessee while its body names sites in Tennessee and Indiana.
Sources: S2
The system effect to watch is whether warehouses increasingly divide work by the certainty of their interfaces. Highly structured flows may favor specialized equipment integrated tightly with storage, conveyors and warehouse software. Less predictable gaps between systems may create room for robots designed to move through human-oriented spaces. That is an inference from the contrast in these reports, not a demonstrated market outcome. The evidence so far favors disciplined task definition over claims that any single robot category has already won the warehouse.
Why it matters
Warehouse buyers can use this comparison to separate an automation capability claim from an operating case. Edeka’s deployment supplies a specific task, environment and rate; Digit’s GXO record supplies evidence of repeated tote handling and a broader expansion, alongside an explicit warning that ROI has not yet been achieved. The near-term decision is likely to be which tasks can be specified, integrated and measured—not whether specialized robots or humanoids are categorically superior.
Sources
- Edeka adds Knapp robots to automate meat picking at German distribution center — Robotics & Automation News ·
- Agility Robotics expands GXO Logistics deal to deploy Digit humanoid in more facilities — Mobile Robot Guide ·