# Automated Scrap Triage

*/Opportunities/Automated_Scrap_Triage*

## Opportunity Overview

**Wedge**: The initial beachhead targets aluminum extrusion facilities and their direct recycling partners. Distinguishing between specific aluminum alloys visually is impossible for humans but highly profitable when sorted correctly, providing fast and concrete proof of ROI. Expansion moves horizontally into copper and brass streams, followed by vertical integration into commodity trading platforms to auto-list the graded scrap inventory.
**Timing**: Edge-compute hardware and multi-modal vision models now process complex, non-uniform visual data in harsh industrial environments without cloud latency. Previously, deploying computer vision on factory floors required massive bespoke datasets, but modern foundational models categorize novel scrap geometries and textures out-of-the-box.
**Why This I C P**: Mid-market industrial metal recyclers and large CNC machining operations face acute labor shortages for hazardous sorting jobs and directly lose margin when high-grade alloys are downgraded. They have both the immediate financial incentive of yield recovery and the operational pain of unfilled headcount to deploy an automated solution.
**Size Of Prize**: There are roughly 35,000 mid-to-large metal fabrication, machining, and industrial recycling facilities in the US. Each absorbs an estimated $80,000 annually in manual scrap handling labor and downgraded material costs due to misclassification, yielding a $2.8B addressable market.
**Gap Narrative**: Industrial recyclers and heavy manufacturers generate mixed scrap streams that require manual sorting to determine metallurgical composition and resale value. Legacy optical sorters process only uniform waste, leaving high-margin mixed industrial scrap reliant on slow, error-prone human visual inspection. An automated, vision-based triage system categorizes and routes scrap at the point of generation to capture lost yield.
**Defensibility**: The system compounds value by aggregating a proprietary dataset of varied, non-uniform scrap materials under diverse lighting and physical conditions, driving model accuracy beyond baseline capabilities. Long-term defensibility relies on workflow lock-in; once the vision system acts as the facility's system of record for inventory valuation and yield reporting, removing it breaks the financial reporting chain.
**Why This Thesis**: A Service-as-Software model delivered via edge-compute appliances matches the need for real-time, localized physical sorting on existing conveyor systems. The physical appliance acts as the ingestion point, while the software layer continuously maps localized scrap streams to live commodity prices without requiring internal IT resources from the buyer.

## Opportunity Linked Thesis

**Thesis**: [Agent](/Theses/Agent)

## Opportunity Linked I C P

**Icp**: [Metal Fabrication Plant](/CompanyTypes/Metal_Fabrication_Plant)

## Opportunity Market Sizing

_Illustrative — target and order-of-magnitude estimate figures, not an achieved track record (this Thing is concept-stage)._

**S A M**: ~$400M-600M (US and EU mid-market high-mix precision fabrication plants)
**S O M**: ~$10M-25M
**T A M**: ~40k global metal fabrication and machining facilities × ~$50k/yr average scrap management cost ≈ ~$2B
**Growth Rate**: ~7-10%/yr, driven by rising alloy commodity prices and increasing margin pressure on scrap recovery
**Paid Comparable Spend**: ~$30k-70k/yr spent on manual floor labor for scrap sorting and lower payout rates for mixed-metal bins

## Opportunity Incumbents

- [SAP Quality Management](/Products/SAP_Quality_Management) — Tool
- [Plex Smart Manufacturing](/Products/Plex_Smart_Manufacturing) — Tool
- [Excel Defect Trackers](/Products/Excel_Defect_Trackers) — Spreadsheet
- [Manual Quality Inspection](/Products/Manual_Quality_Inspection) — Service
- [ETQ Reliance QMS](/Products/ETQ_Reliance_QMS) — Tool
- [Custom Python Scripts](/Products/Custom_Python_Scripts) — DIY
- [Cognex Vision Systems](/Products/Cognex_Vision_Systems) — Tool

## Opportunity Win Conditions

**Kill Thresholds**:
- Alloy misidentification rate > 4 percent after 30 days
- User override and bypass rate > 25 percent of daily throws
- Increase in scrap buyer payout < 10 percent within 90 days
- Hardware and installation CAC > $8,000 per facility
**Leading Metrics**:
- Time from installation to first automated bin sort
- Percentage of daily offcut volume processed through the system
- Alloy misidentification rate per 1,000 scans
- Scrap buyer purity dispute frequency
- Weekly hours spent on manual system recalibration
**What Proves Right**: Facilities install the sensor system above scrap bins and sort at least 80 percent of daily offcuts without manual logging. They negotiate a 15 to 20 percent higher payout rate from their scrap buyers within the first two billing cycles because the alloy purity is guaranteed. Cohorts retain at over 90 percent after 90 days at a $2,500 monthly subscription tier because the system covers its own cost in yield.
**What Proves Wrong**: Floor workers ignore the automated bin routing and continue tossing offcuts into mixed piles due to speed pressures. The triage system misidentifies alloys at a rate higher than 5 percent, causing scrap buyers to reject the purity claims and downgrade the payout tier. The system requires more than one hour of weekly recalibration by a floor manager, offsetting the labor savings.

## Opportunity Build Profile

**Hardest Part**: Achieving sub-second classification accuracy on highly occluded, deformed, or dirty materials under variable factory lighting. The system must reliably distinguish between visually similar scrap grades without triggering false positives that contaminate downstream recycling batches.
**Min Viable Scope**: Deliver passive computer vision grading for a single material stream on one conveyor belt to trigger dashboard alerts on contamination. Deliberately exclude multi-material mixed sorting, hardware robotic ejection mechanisms, and upstream predictive yield analytics.
**Cold Start Problem**: The model requires thousands of annotated images of specific scrap variants in realistic, messy environments before it operates reliably. Break this by deploying passive edge cameras at a single design partner's facility to collect raw video of the scrap stream for offline, expert-led labeling prior to launching the active triage engine.
**Time To First Value**: 3 to 4 weeks, gated by physical edge camera installation, lighting calibration, and localized model fine-tuning
**Data Moat Available**: true
**Technical Difficulty**: High

## Neighborhood

### Where the gap lives

- [Quality Control Inspector](/JobTypes/Quality_Control_Inspector) — latent gap · JobTypes
- [Measurement System Variance](/Metrics/Measurement_System_Variance) — latent gap · Metrics
- [Production and Processing](/Knowledge/Production_and_Processing) — latent gap · Knowledge

### Incumbent in

- [Excel Defect Logs](/Products/Excel_Defect_Logs) — incumbent in · Products
- [Bespoke Python Scripts](/Products/Bespoke_Python_Scripts) — incumbent in · Products
- [Cognex Machine Vision](/Products/Cognex_Machine_Vision) — incumbent in · Products
- [SAP Quality Management](/Products/SAP_Quality_Management) — incumbent in · Products
- [ETQ Reliance QMS](/Products/ETQ_Reliance_QMS) — incumbent in · Products
- [Manual Quality Inspection](/Products/Manual_Quality_Inspection) — incumbent in · Products
- [Plex Smart Manufacturing](/Products/Plex_Smart_Manufacturing) — incumbent in · Products

### Applies thesis

- [Metal Fabrication Plant](/CompanyTypes/Metal_Fabrication_Plant) — applies thesis · CompanyTypes

### Embodies

- [Agent](/Theses/Agent) — embodies · Theses

### Similar Opportunities

- [Sort Node Analyzer](/CompanyTypes/Waste_Management_and_Remediation_Services_Provider/Opportunities/Sort_Node_Analyzer) — similar · Opportunities
- [OpticSort Labs](/Opportunities/OpticSort_Labs) — similar · Opportunities
- [Machine Vision Baling Inspection](/Opportunities/Machine_Vision_Baling_Inspection) — similar · Opportunities
- [Spectroscopic Grading for Metal Foundries](/Opportunities/Spectroscopic_Grading_for_Metal_Foundries) — similar · Opportunities
- [Yield Optimization Service](/Occupations/Production_Occupations/Opportunities/Yield_Optimization_Service) — similar · Opportunities
- [Autonomous Baling Line Vision](/Opportunities/Autonomous_Baling_Line_Vision) — similar · Opportunities
- [Visual Defect Detection](/Opportunities/Visual_Defect_Detection) — similar · Opportunities
- [Feedstock Arbitrage Broker](/Opportunities/Feedstock_Arbitrage_Broker) — similar · Opportunities
- [Zero-Day Defect Sentinel](/Opportunities/Zero-Day_Defect_Sentinel) — similar · Opportunities
- [Computer Vision Scrap Detection](/Opportunities/Computer_Vision_Scrap_Detection) — similar · Opportunities
- [Optical QA as a Service](/Opportunities/Optical_QA_as_a_Service) — similar · Opportunities
- [AI Scrap Mitigation for Motorsports](/Opportunities/AI_Scrap_Mitigation_for_Motorsports) — similar · Opportunities
- [Algorithmic CNC Scrap Reduction](/Opportunities/Algorithmic_CNC_Scrap_Reduction) — similar · Opportunities
- [Predictive Material Yield Management](/Opportunities/Predictive_Material_Yield_Management) — similar · Opportunities
- [Automated Quality Control](/Opportunities/Automated_Quality_Control) — similar · Opportunities
- [Automated Teardown Analysis for Manufacturing](/Opportunities/Automated_Teardown_Analysis_for_Manufacturing) — similar · Opportunities
- [AI Supplier Validation](/Skills/Quality_Control_Analysis/Opportunities/AI_Supplier_Validation) — similar · Opportunities
- [Offcut Optimization for Furniture Makers](/Opportunities/Offcut_Optimization_for_Furniture_Makers) — similar · Opportunities
- [Automated Visual Inspection](/Opportunities/Automated_Visual_Inspection) — similar · Opportunities
- [Beneficiation Yield Optimizer](/Opportunities/Beneficiation_Yield_Optimizer) — similar · Opportunities