A potato processor can be too thorough. Set the peeler aggressively enough to remove every trace of skin and it may also remove flesh that could have become a saleable product. Polysense’s potato-processing documentation identifies that safety margin as a target for its AI: measure the peel remaining and keep adjusting the machine as the raw material changes.[1]
The Belgian company announced its first US office, in Chicago, on 22 September. Co-founder Jarne Bogaert is relocating to lead it, with the team focused on the US and Canada. The expansion follows a $10.7 million seed round in July. The business is selling a connection between product inspection and the controls of an operating food factory.[2]
Quality can change while the recipe stays put
Polysense describes raw ingredients whose moisture, density and other characteristics vary between batches. A fixed production setting can continue applying the same treatment after the input has changed. Its system uses images from the line to identify quality changes, combines measurements in a common platform and can adjust machine parameters automatically.[3]
Peeling makes that relationship visible. The target is an acceptable peeled surface while retaining as much usable potato as possible. Polysense says it measures remaining peel and over-peeling, then can adjust steam or abrasion settings. The same potato-processing offering includes monitoring cut dimensions and breakage, and changing sorter thresholds to reduce rejection of acceptable product.[1]
These are different decisions with a common economic consequence. A severe peeling setting may satisfy a quality requirement while sacrificing material. An overly aggressive sorter can make the accepted stream look cleaner while sending valuable food into the rejected stream. Seeing only the quality of what survives would miss the cost of producing it.
The measurement has to reach the machinery
Polysense separates inspection, called Qualify, from automatic intervention, called AutoControl. The former supplies quality measurements for operators; the latter uses them to change the machine’s setpoints. Its documentation says AutoControl connects through existing industrial control systems, including the OPC-UA and MQTT communication protocols. Product-specific quality criteria determine what the system measures when a line changes over.[4]
The named Agristo case concerns inspection on a French-fry line. Polysense describes replacing intermittent manual checks with continuous assessments of length, defects, colour and shape, feeding a dashboard integrated with Agristo’s production and enterprise systems. The account reports a shorter feedback loop and more consistent inspection, without publishing a numerical waste reduction.[5]
Feedback time has a physical cost. If material is being lost at a steady rate, halving the correction delay halves the amount lost during that delay. That illustrative relationship explains why earlier information can matter even before a manufacturer delegates the adjustment.
What the peeling result measures
In its July funding announcement, Polysense reported that an unnamed European potato processor had reduced peeling time by 45% using real-time peel monitoring and automatic adjustments. The release does not identify that customer as Agristo or provide the baseline duration and trial period. The measure is peeling time; it cannot be read as a 45% increase in factory output or a 45% reduction in waste.[3]
For the manufacturer, the eventual return depends on what the change preserves: usable material, capacity at a busy process stage, operator time or energy. These benefits need different measurements. More good potato from the same incoming load concerns yield; a shorter process concerns duration. A profitable installation may improve both, but one percentage cannot stand in for the other.
Polysense’s Chicago announcement gives a concrete purpose for the new office: a local team close enough to understand customer lines, with delivery linked to its Belgian operation. That matters for a product whose useful action depends on the equipment, ingredients and quality requirements at each factory.[2]
A responsive peeler can ease off when the next batch needs less treatment. Its value would leave the factory in the product: good potato that continued down the line.
Sources & further reading
- AI Quality Control for Potato Processing Polysense · Accessed 19 Oct 2026
- Polysense opens its first US office, in Chicago Polysense · 22 Sep 2026 · Accessed 19 Oct 2026
- Polysense raises $10.7 million to shape the food factory of the future with AI and computer vision Polysense · 8 Jul 2026 · Accessed 19 Oct 2026
- Polysense AutoControl — Automated process control based on real-time data Polysense · Accessed 19 Oct 2026
- How Agristo is saving waste, time and money with Polysense Qualify Polysense · 29 May 2026 · Accessed 19 Oct 2026