data_entry_ops · services · workflow
Kantar Worldpanel uses Databricks and GPT-4 to generate 120,000 training pairs at 94% accuracy
Kantar Worldpanel's legacy systems were inflexible, resource-intensive, and required an outdated programming skillset, blocking their ability to experiment with modern AI/ML models and scale their data insights business.
How it works
Common implementation structure
How this type of workflow is generally built, generalized across documented cases — not tied to any one vendor's stack. Click any stage to read what happens there. Specific products that implement these stages appear in “Tools commonly seen” below.
Stage 1 · Receipt-to-barcode matching need
Pairing receipt line descriptions and product barcode names is the upstream use case that starts the workflow.
Tools used
DatabricksMLflowMosaic AI Vector SearchUnity CatalogDatabricks MarketplaceLlamaMistralGPT-4GPT-3.5
Outcome
Kantar Worldpanel automatically generated a training dataset of about 120,000 pairs of receipt descriptions and barcode names at 94% accuracy in a couple of hours, freeing manual coding teams and engineering resources for higher-value work.
Results
Time savedin just a couple of hours
Volume94%
Cost replacedmore cost-effective but more performant
Grounding & classification
Source type: vendor customer story
34 fields verified against source quotes.
content generationdata extractionpredictive analyticsproduct catalogreceipthuman review describedmetric backednamed customerproduction runtime claimedsource backedtools describedworkflow describedprofessional servicesaccuracy improvementcost reductionemployee productivitytime savedvendor customer storyback office opsdata entry opsdocument to record