KPM Analytics SiftAI — AI foreign-material detection on US beef-trim lines (undisclosed customer names)

NA-US ("major U.S. beef harvest and secondary processing facilities" — customer names not publicly disclosed)

Content

KPM Analytics (US-based) sells SiftAI Foreign Material (FM) Detection — an AI vision system that detects foreign material (plastic, wood, styrofoam, rubber, cardboard) on beef-trim lines. SiftAI complements metal detection and X-ray; the AI vision layer is trained to distinguish product from FM even when visually similar, which metal detection and X-ray cannot do.

Deployment evidence:

• Per KPM Analytics technical-podcast guest Vernon W. Stout (Technical Operations Director, Vision Inspection, KPM Analytics), SiftAI FM is “deployed within major U.S. beef harvest and secondary processing facilities” — plural, but specific customer names are not publicly disclosed in the podcast
• Vendor product page describes SiftAI FM as deployed in commercial operations; customer list confidential
• The technique is complementary to metal detection + X-ray, not a replacement

Why this is a uniquely-thin unit. SiftAI is a confirmed deployed NA AI vision system in beef processing, but the customer names are not public. This is the inverse of Cargill CarVe (publicly named, in-house) and TOMRA / Key Technology (publicly named, sold to multiple customers). SiftAI is sold to a closed set of large customers whose names are protected — common in B2B industrial-AI but worth surfacing because it makes the knowledge base’s “named customer” pattern uneven.

What this unit is doing in the taxonomy

Anchors the processing × computer vision × food-safety cell at a sub-cell that other units don’t occupy: B2B-sold AI vision with confidential customer list (vs Cargill CarVe = in-house, TOMRA/Key Tech = public multi-customer). Sister units:

• Cargill CarVe (cargill-carve-meat-processing.md) — beef, line-side CV, in-house proprietary
• Tomra NA (tomra-na-food-sorting.md) — multi-customer optical sorting with foreign-material detection capability
• Key Technology NA (key-technology-na-food-sorting.md) — US-HQ multi-customer optical sorting

Why it matters for talks

• The “complements metal detection + X-ray” framing is the load-bearing claim. Metal detection and X-ray are the legacy food-safety inspection modalities in beef processing; SiftAI fills a gap neither could cover (visual similarity FM such as clear plastic in clear packaging). Worth distinguishing from CarVe (yield-aimed CV) and TOMRA/Key Tech (defect-removal CV).
• The unpublished customer list is itself the structural observation. Many B2B industrial-AI deployments happen without public customer disclosure; this is a feature of the procurement pattern, not a knowledge-base gap.
• Food-safety framing is the deepest cell. Vision-AI in food safety is rare in NA; most NA food-AI vendor narratives emphasise yield or efficiency, not safety/regulatory. SiftAI sits in the safety cell.

Critical context

• Customer names not disclosed — vendor and customer confidentiality. The deployment footprint is “major” beef processors, plural, but not enumerated.
• The technique class is vision-based with ML distinguishing FM from product; ML model architecture is not publicly described.
• The deployment scale (number of lines / plants) is not disclosed.
• Comparison to metal detection + X-ray is by KPM; independent third-party comparison not surfaced.
• Whether SiftAI certifies against USDA-FSIS regulatory standards is not surfaced.