AGCO Plans 2027 Digit Pilot at Its Hesston Factory

AGCO plans to test Agility Robotics' Digit humanoid at its Hesston, Kansas, factory in 2027. The agricultural-equipment manufacturer announced a signed commercial agreement on October 20, with initial work covering parts sorting, stacking and tote placement for storage. The companies have identified a customer site and handling applications; they have not reported an operating Hesston fleet. [1]

For humanoid robotics, the interesting test is the connection between the robot and the surrounding production process. Moving a container can look straightforward in isolation. At a factory, useful automation must also deliver the correct material, to the correct destination, when the next operation needs it. Hesston gives that question a specific industrial setting, while leaving the result for the pilot to establish.

The Hesston agreement

AGCO makes agricultural machinery and precision-agriculture technology through brands including Fendt, Massey Ferguson, PTx and Valtra. This announcement concerns the factory that builds equipment, rather than a humanoid working on a farm. AGCO describes reducing physical strain and improving manufacturing flexibility as intended benefits. It has not supplied measured improvements in ergonomics, output or cost from this pilot. [1]

The word commercial establishes an agreement between the parties. It does not, by itself, establish the number of machines purchased, the payment structure or completed deployment. Those terms are undisclosed. A signed pilot can be a meaningful procurement step without supplying evidence that the eventual workflow meets its production requirements.

Biped's assessment is that the named applications make this more useful than a general statement of interest in humanoids. They give readers something concrete to follow. Subsequent reporting can examine whether the trial began, which handling jobs reached acceptance and whether AGCO elected to expand them. None of those later decisions should be assumed from the agreement.

Under the hood: handling work and product specifications

The announcement names Digit without specifying its generation. That distinction matters because Agility's upcoming Digit 5 has a separate hardware and availability roadmap. Its September launch described a repeated-lift capacity of 22.7 kilograms, a reach of 2.2 meters and interchangeable wrist tools. These are manufacturer specifications for Digit 5, not a disclosed configuration for AGCO. [2]

Agility currently warns that Digit 5 remains in development, with specifications subject to change and some safety features unfinished. A future product's advertised capability therefore needs to stay separate from a customer's agreed operating conditions. The timing of the two plans overlapping in 2027 does not establish that they use identical hardware. [3]

For a handling trial, the practical specification is the whole task. Container geometry, load distribution, pickup height and destination can affect how demanding the work is. A maximum payload alone cannot answer whether a robot can handle a particular part or storage arrangement at the required pace. Hesston's application details will be more informative than transferring a product-sheet number into a factory forecast.

The three application labels also hide different problems. Sorting requires a way to distinguish the relevant parts and recognize an incorrect selection. Stacking adds placement and stability requirements: completing one successful placement says little about the consistency of a full stack. Tote storage introduces the handoff to whatever system or person retrieves the container next. The announcement does not describe how these checks will be implemented.

Those differences suggest a more useful reporting format than one aggregate success rate. Results could identify the task, the accepted item types and the conditions of the trial, then explain which exceptions required help. A narrow application performed consistently may justify continued use even if broader manipulation remains unresolved. Conversely, a demonstration across several objects does not establish that every supported variation is ready for production.

Agility's deployment playbook

Agility's August account of its Customer Acceleration Program describes controlled workflow testing at its own facility, followed by validation at the customer site and measurement of operational impact. It says its process considers object weights, layouts, obstacles, network connectivity and the connection to its Arc fleet software. This is the supplier's general process; the October agreement does not identify which stages AGCO has completed. [4]

The same account acknowledges that a humanoid is not always the appropriate answer. A fixed arm or an autonomous mobile robot may better fit some applications. That is a useful starting point for evaluating Hesston: walking and manipulation should solve a particular integration problem, rather than make a familiar task more technically elaborate.

There is a manufacturing precedent in Agility's customer history. Its February 19 announcement with Toyota Motor Manufacturing Canada described a Robots-as-a-Service agreement following a successful pilot. That provides an example of a reported transition from evaluation to a continuing commercial arrangement. It does not establish that AGCO will follow the same path, buy the same service or obtain the same results. [5]

What the pilot needs to measure

A useful evaluation would distinguish attempted moves from accepted work. If a tote arrives late, a part is mis-sorted or a person must correct placement, the robot's motion may have succeeded while the production task has not. Reporting accepted output alongside assistance and recovery time would make the result easier to interpret. These are Biped's proposed evaluation questions, not disclosed AGCO acceptance criteria.

Charging is another part of that accounting. On October 15, Agility identified Nyobolt as Digit 5's battery-technology partner and described a planned 90-minute runtime with a nine-minute recharge. The companies expect this to support more than 20 hours of work per day. Those are forward-looking product claims, not measured Hesston availability. [6]

Even if a charging target is achieved, total useful time also depends on maintenance, waiting for material, intervention and whether the site has enough suitable work. An economic comparison should include those interruptions and the cost of integration. With no disclosed AGCO price or throughput, a savings percentage or payback period would be speculation.

Workforce effects need similarly specific evidence. Taking over a strenuous movement can change a person's job without eliminating the entire role. Someone may still replenish materials, inspect output or handle exceptions. Measuring how assistance and responsibilities change would help distinguish reduced physical exposure from a simple transfer of work elsewhere. The announcement's intended employee benefits do not establish a staffing change or a measured reduction in injuries.

Safety belongs to the application

Agility and FORT Robotics announced an October 1 memorandum of understanding covering safety hardware, communications and connections to external safety systems. Their Digit 5 plans include a pendant for monitoring and manual override, plus an offboard interface intended to extend protection beyond the robot. This describes development work, not approval of a particular Hesston layout. [7]

The distinction becomes practical when parts, people and moving machinery share a work area. An evaluation needs to establish who can stop the process, how abnormal situations are handled and what must happen before work resumes. AGCO has not disclosed those arrangements. General statements about cooperative safety should not be read as evidence of unattended operation or unrestricted proximity to workers.

What's coming next

The first milestone is confirmation that the 2027 trial has started, with its configuration and scope identified. The next is evidence that the selected jobs work reliably over a stated period. Expansion to another application, shift or factory would then offer a different signal: a customer's decision to commit further after evaluation. Keeping those steps separate makes real progress easier to recognize.

Frequently asked questions

Is Digit already working at AGCO's Hesston factory?

The companies announced a pilot expected to begin in 2027. They have not reported an operating Hesston fleet or published performance results.

Will AGCO use Digit 5?

The announcement names Digit without specifying its generation. Digit 5's specifications and availability plans should not be treated as AGCO's disclosed configuration.

What work is planned for the pilot?

The initial applications are parts sorting, stacking and tote placement for storage. The companies have not disclosed robot count, task throughput or supervision arrangements.

Does the photograph show the AGCO pilot?

No. It is an Agility press-kit photograph of Digit holding a tote. It is product context, not a photograph of the future Hesston installation.

A factory test with a defined purpose

AGCO has supplied a place and a set of jobs for its humanoid evaluation. The strongest follow-through would connect that plan to accepted output, practical supervision and a decision about continued use. Those disclosures would help manufacturers judge where this kind of machine earns its place alongside existing automation.