Case picking automation fails when nobody schedules the human

Vecna Robotics built its first warehouse platform around one question: what can the robot do. Then the company took that platform into case picking, and the question stopped being enough. A robot could reach the right pallet slot on schedule, and the pick still failed if the person who had to lift the case was two aisles away.

That is the problem a Vecna speaker lays out in a recent MarketScale Studio segment on CaseFlow, the company's case picking system. Case picking means pulling cartons off pallets in racking and building a mixed pallet for an order, and in most warehouses a person does the lifting. Vecna's answer was to stop treating the robot as the whole product and to schedule the human with the same software.

The timing matters because CaseFlow is no longer a pilot story. Robotics 24/7 wrote in October 2026 that Vecna reported a two times throughput gain and up to 90% less worker travel at a GEODIS deployment, and in a March 31, 2026 press release Vecna announced CaseFlow Voice, a hands-free version for cold storage it demonstrated at MODEX 2026 in Atlanta. The design choices the speaker describes are the ones operators are now buying.

A robot at the right slot is useless if the picker is two aisles away

Case pickers spend as much time walking a distribution center as selecting product, Robotics 24/7 wrote in October 2026. That is the scale of the problem behind the speaker's point that a robot reaching the slot is only half the pick. The other half is a worker arriving at the same slot at the same moment, and nothing in a robot-only dispatcher accounts for where that worker is standing.

Vecna's software positions robots near upcoming pick locations and assigns a nearby worker to the next task, according to the same report. The system weighs the locations of both workers and robots rather than optimizing robot movement alone. For an operator, that is the difference between a fleet that drives efficiently and a fleet that picks efficiently.

In its March 2026 press release, Vecna said CaseFlow orchestrates pallet-handling robots and human workers in real time, balancing robot availability, worker location and order priorities. Workers stay inside smaller, dynamic zones and pick, while autonomous mobile robots (AMRs, self-driving vehicles that find their own path through a facility without tape or wires) carry the pallets between zones. The walk shrinks to a few steps inside a zone; the haul across the building belongs to the robot.

The company says in the interview that getting there took effort, because orchestrating the robots was the part it already knew, and the humans then had to be orchestrated as well and told what to do.

Robotics 24/7 describes that application as wearable devices for worker guidance. The wearable tells the worker where to go and what to load. The robot's route and the worker's route come out of one scheduler, which is what keeps the two from missing each other at the slot.

The robot was the product until case picking put a person in the loop

Vecna started with the platform as a robot and asked what the robot could do, the company says in the interview. Case picking forced a second question, because the robots reaching the right locations was only part of the task; the humans had to meet them there at the right time to do the work. That is the point where a vehicle becomes a workflow, and where the software has to know what the order needs, not only where the robot is.

The order system in question is the warehouse management system (WMS), the software that holds inventory locations and customer orders and generates the pick lists workers follow. On its own site, Vecna says CaseFlow integrates with the WMS and other operational systems to build a real-time view of warehouse demand. In its March 2026 press release, the company said its systems connect to existing WMS, enterprise resource planning (ERP, the finance and inventory system of record) and manufacturing execution platforms, and need no fixed infrastructure.

The practical meaning for an operations lead is that the integration is the project. The robots are hardware that drives; the value sits in whether the software can read the order feed, split it into robot stops and worker assignments, and keep both current as orders change through the shift. That is why the company frames the product as a workflow rather than a vehicle.

The alternative Vecna measured against costs more and digs into the floor

The speaker names one class of competing technology on the recording and sets it aside on cost and disruption.

The speaker does not expand the acronym, and neither Vecna nor any public comparison prices the two approaches for the same site. What Vecna says it set out to build instead was something easy to introduce into an existing building without heavy infrastructure cost. On its own site, the company lists what it designed against: fixed infrastructure that requires facility modification, large upfront capital with long timelines, one-size-fits-all automation that forces the operation to adapt to the technology, and siloed systems where workers, equipment and software each run on their own with no shared view.

The nearest vendor comparison comes from a different approach. Prime Robotics, which sells goods-to-person AMRs, said in its own product material in October 2023 that AMRs need minimal facility modification compared with fixed automation such as conveyors and sortation. Goods-to-person means the inventory travels to a stationary worker at a workstation, often retrieved from an automated storage and retrieval system (AS/RS), the racking, crane or shuttle hardware that stores and fetches totes or pallets mechanically.

Prime Robotics says on its own site that goods-to-person lifts productivity up to six times; Vecna says on its site it delivers two times or more with no fixed infrastructure. That gap is the honest complication in Vecna's thesis, because a higher multiple is on offer from the model it calls intrusive. Prime's own page adds that its heavier goods-to-person robots can carry full pallets for case picks, while lighter personal-assist robots that follow a walking picker generally support only each-picking, the selection of single items rather than whole cases.

Vecna's counter is time rather than the multiple. On its site the company puts deployment at four weeks or less, and the speaker's argument is that an operator with an existing building and an existing workforce cannot absorb the floor work a fixed system requires. Each vendor measures its multiple on its own sites and its own workloads, so an operator weighing the two is comparing two companies' own claims, not a shared test.

An exception the worker cannot complete has to have somewhere to go

Because case picking was Vecna's first product that put a human inside the loop, the company says it had to ask what the worker needs to do at every step and leave none of it out. Exceptions were the test case.

The requirement is simple to state: whatever stops a pick, the worker must be able to do the job or report back that they cannot, so the robot and the order do not hang on a task nobody can finish. The speaker does not say where a reported exception goes next, whether the line is rerouted to another worker or returned to the WMS as a short.

CaseFlow Voice lets a gloved cold-storage picker take instructions and confirm a task by speaking, Vecna announced in March 2026. The press release says cold storage defeats screens and scanners because gloves, condensation and limited dexterity get in the way, so the voice version embeds Lucas Systems' Jennifer voice direction into the picking workflow while keeping a handheld interface as an option. Food Logistics reported the launch on April 1, 2026 and noted the same real-time balancing of robot availability, worker location and order priority.

The exception problem turns out to be an interface problem. A report-back that requires a bare hand on a cold screen is a report-back that does not happen. In the release, Karl Iagnemma, Vecna's CEO, said the system has already cut training time by up to 50%; the release puts Jennifer's voice options at 37 languages, and Lucas Systems chief growth officer Chris Simchick said in the same release that Jennifer has powered more than 112 billion picks.

Flow means never sending every robot to the same aisle

The name was deliberate, the company says: CaseFlow, rather than case picking. The failure it describes is a fleet dispatched purely by order priority, where several robots converge on one hot aisle and block each other while workers in every other zone wait.

Vecna's own material from March 2025 calls the method dynamic, directed zone picking, and put next to the Robotics 24/7 description the mechanism follows from the zone. A worker is assigned a stretch of aisles rather than an order; the robot carrying the order's pallet is positioned near its next pick location, and the nearest worker is sent to meet it there. Because the pallet moves through zones and each zone has its own picker, no single slot draws the whole fleet.

On its own site, Vecna says the system rebalances resources and workflows in real time to raise throughput and lower cost per case, and that worker zones are redrawn through the day. For an operator, the test of that claim is visible from the floor: a dispatcher that is working spreads robots across the building, and one that is not leaves a queue at the busiest bay.

Vecna's tugger moves up to three carts at once; its forklift handles heavier palletized loads on the same software, according to its site. The company lists three form factors under one orchestration layer: the co-bot pallet jack for high-volume case picking in retail, grocery and 3PL distribution centers (3PL, an outsourced logistics provider running a warehouse for other companies), the autonomous tugger for multi-load transport, and the autonomous forklift for food and beverage and manufacturing sites with heavier pallets. The flow logic does not change when the vehicle does, which is the point of designing the software above the hardware.

Research says who leads the pick still changes the outcome

Erasmus researchers ran 60 pickers through a 300-location test warehouse to see whether human or robot should lead. Alexandros Pasparakis, Jelle de Vries and René de Koster described the experiment, built in the Netherlands, in a September 2026 article on Interlake Mecalux's site. A cobot is a collaborative robot, one built to work beside a person rather than inside a fenced cell, and participants either followed a cobot to the slot the WMS indicated or led the route themselves while the robot followed, taking instructions by voice or by pick-to-light, the lights at a slot that show a picker what to take.

The researchers write that recent research argues relieving pickers of travel can make cobot-assisted picking more productive than conventional order fulfillment. They raise two variables: humans and AMRs often alternate between leader and follower roles depending on the task, and human behavior may be affected by prevention focus, a psychological measure of how vigilant a person is in carrying out a task. The study used shelving for small, low-volume items rather than cases, and the researchers frame who should lead as the open question, so it bounds Vecna's thesis rather than refuting it.

CaseFlow sits between the two modes the researchers tested. The software assigns both the robot's stop and the worker's destination, and the worker receives the instruction on a wearable, according to Robotics 24/7. Iagnemma told the publication in October 2026 that the system is human-robot collaborative by design, with the fleet introduced to raise picker productivity rather than to remove the picker, which is Vecna's answer to who leads: the orchestration layer does.

The GEODIS site is the benchmark a pilot should be held to

GEODIS logged seven powered-equipment incidents in 2022 and two during the two-year CaseFlow rollout, Robotics 24/7 reported.

In its own material from March 2025, Vecna said the Indianapolis pilot more than doubled workflow productivity with no reported safety issues, and that payback came in under 12 months through robots-as-a-service pricing, a subscription in which the operator pays for robot capacity rather than buying the vehicles. The same material says the rollout required no disruption to existing operations, which is the claim behind the speaker's point about introducing a system into a building that already runs.

The questions the interview hands an operator are concrete. Does the worker app let a picker report a pick they cannot complete, in gloves, in the cold. Does the dispatcher spread robots across zones or send them to the hottest aisle. Which WMS fields does the software consume, and what happens to a line that cannot be filled.

Pasparakis, de Vries and de Koster give the operator one more thing to measure in a pilot: whether pickers who follow a robot behave differently from pickers who lead one. CaseFlow puts the scheduler in charge of both, which is where that difference would show up.