Quality assurance · Production

Tractable's Auto Inspector speeds and improves damage assessment for LKQ's recycled auto parts

The problem

LKQ's inventory management process was manual, repetitive, and time-consuming at huge scale — around 800,000 used autos a year across 165+ locations, with well over 100 recyclable parts to tag and inventory per car — and prior AI damage-assessment solutions LKQ evaluated weren't accurate enough to deploy in live processes.

First attempt

LKQ's VP reviewed several other AI damage-assessment solutions before Tractable, but none were accurate enough to deploy in their live processes.

Workflow diagram · grounded in source
1
Vehicle intake for assessment
trigger
“As each end-of-life vehicle comes in, it must be fully damage assessed for its value”
2
Computer vision damage assessment
ai_action
“Tractable’s Auto Inspector tool uses computer vision to assess the specific damage on a vehicle and then determines which parts can be recycled and reused.”
3
Model trained on damage examples
ai_action
“The AI is trained on millions of examples of auto damage, and can perform on a level with human intelligence, increasing the speed, consistency and accuracy of decisions.”
4
Calibration to LKQ standards
validation
“Tractable worked closely with the LKQ team to ensure that the Auto Inspector tool was calibrated to LKQ’s industry standards regarding accurate damage assessment.”
5
Tagging and inventorying parts
output
“all its recyclable and reusable parts – that could reach well over 100 items for each car – are tagged and inventoried”
Reported outcome

Tractable's Auto Inspector, powered by computer vision trained on millions of damage examples, returned very accurate multi-part damage assessments in about three seconds during testing, and the tool was calibrated with LKQ's team to move from initial testing to production at scale.

Reported metrics
Vehicles processed annually800,000
Locations165
Recyclable parts per vehiclewell over 100 items
Damage assessment turnaroundabout three seconds
Reported stack
Auto Inspector
Source
https://tractable.ai/case-studies/lkq/
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

Tractable's Auto Inspector, powered by computer vision trained on millions of damage examples, returned very accurate multi-part damage assessments in about three seconds during testing, and the tool was calibrated wi…

What tools did this team use?

Auto Inspector.

What results were reported?

Vehicles processed annually: 800,000; Locations: 165; Recyclable parts per vehicle: well over 100 items; Damage assessment turnaround: about three seconds (source-reported, not independently verified).

What failed first in this deployment?

LKQ's VP reviewed several other AI damage-assessment solutions before Tractable, but none were accurate enough to deploy in their live processes.

How is this quality assurance AI workflow structured?

Vehicle intake for assessment → Computer vision damage assessment → Model trained on damage examples → Calibration to LKQ standards → Tagging and inventorying parts.

Related quality assurance cases