Client context
An automotive manufacturer with a large global field population and warranty/technician data spread across claims, dealer and telematics systems.
Problem size
Defect patterns surfaced in warranty claims and technician notes weeks before anyone connected them to a specific build, supplier lot or shift — by then thousands of units were already in the field.
What KEPLER did
- Built an NLP/LLM pipeline reading technician verbatims, DTC codes and warranty claims
- Resolved signals against build genealogy (VIN → build → supplier lot → shift)
- Detected emerging defect signatures and triggered root-cause investigation early
- Surfaced live defect-emergence views in Power BI for quality and warranty teams
Results
Defects caught weeks earlier, shrinking field exposure and speeding root-cause investigation.
Anonymized client engagement; identifying details withheld for confidentiality.