Winnow — UK-hq AI food-waste tracking with substantial NA contract-feeder footprint

NA-US (Winnow deployed at NA contract feeders; EU origin, NA operations); UK (HQ); 94 countries globally

Content

Winnow is a UK-headquartered AI food-waste tracking vendor whose Winnow Vision product uses a camera mounted above the bin plus an integrated scale to automatically identify what is being thrown away in a commercial kitchen. The product is deployed across 3,500+ sites in 94 countries per the company’s FAQ.

Winnow Vision — what it does:

• Camera identifies the food item being discarded (trained on common kitchen items)
• Integrated scale records weight
• System logs item + cost + waste reason over time
• Chefs get analytics dashboards for waste patterns; the vendor claims kitchens using Winnow cut food waste by ~50%

NA footprint evidence (specific):

• ISS — Danish global contract-feeder (GBS division); Andy Harris, Head of Group Business Development at ISS, quoted endorsing Winnow on the FAQ. ISS has substantial NA foodservice operations.
• Elior NA — French-origin contract-feeder with NA presence; named as a Winnow customer reference, but Elior NA also announced a Leanpath deployment. This is interesting structural evidence: major contract feeders are using both Winnow and Leanpath across their portfolios — likely site-by-site procurement — not exclusive one-vendor relationships.
• Winnow’s case-study and customer pages list global deployments but the FAQ isolates ISS as the named large contract-feeder customer.

Why this is the companion unit to Leanpath:

• Leanpath is the long-tenured US-internal food-waste AI vendor (founded early 2000s, Portland OR HQ, US Foods partnership).
• Winnow is the UK-international food-waste AI vendor that has gained NA share via global contract-feeder procurement (ISS, Elior).
• Both vendors converge on the same headline figure (~50% reduction). The convergence is itself a structural observation worth surfacing.

What this unit is doing in the taxonomy

Anchors the waste-and-recovery × computer vision × supply-chain-efficiency cell with the international-vendor-with-NA-presence pattern. Sister unit to:

• Leanpath (leanpath-food-waste-tracker.md) — US-internal vendor; Winnow is the international counterpart.

Why it matters for talks

• The “both Leanpath and Winnow at Elior NA” observation is the structural insight. Major contract feeders run multi-vendor product stacks — different brands, regions, and operational units pick different vendors. There is no single-vendor lock-in at this layer of the food-waste AI stack. Worth distinguishing from equipment-vendor lock-in patterns elsewhere in the field (e.g. John Deere See & Spray, which is equipment-bundled).
• The 3,500+ site count matters at a different level than Leanpath’s “thousands of kitchens” claim. Both vendors describe thousands of installations — the order-of-magnitude alignment across competitors is a market-structure observation, not just a marketing coincidence.
• Winnow’s data flows across borders. UK-headquartered vendor with NA customers means kitchen-floor image data may flow across the Atlantic. Data sovereignty is a working concern that did not arise for Leanpath (US-HQ, US-deployed).
• Both vendors converge on the same 50% reduction claim. Worth flagging because the convergent figure is itself a methodology question — are they measuring the same thing? Are they excluding the same waste streams?

Critical context

• All impact metrics are vendor-reported (50% reduction cited in Winnow, Leanpath, and ReFED’s coverage).
• Customer roster for NA sites is less directly enumerated than Winnow’s EU customer pages; ISS is named on the FAQ, Elior NA appears in both vendor lists.
• “3,500+ sites in 94 countries” is a Winnow FAQ claim; site count not independently audited.
• Winnow and Leanpath are not one-vendor-replaces-the-other — major contract feeders deploy both. Worth not overstating either vendor’s exclusivity in NA.
• Global (UK-HQ) data flows for NA customer data have not been publicly characterised by Winnow — what is captured, where stored, who can access aggregated patterns across sites.