ecommerce_ops · ecommerce · workflow
Mercari AI Assist: GPT-4 and GPT-3.5-turbo power seller listing title suggestions in production
Processing and understanding unstructured user-generated listing text was difficult for Mercari's team, as distilling key information and identifying what made listings sell quickly was complex given the varied writing styles of sellers.
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 · Domain experts define title criteria
Teams with diverse domain expertise assist in defining what makes a good title for a Mercari listing.
Tools used
GPT-4GPT-3.5-turbo
Outcome
Mercari shipped Mercari AI Assist with a two-stage LLM pipeline—GPT-4 for offline key-attribute extraction and GPT-3.5-turbo for real-time inference—achieving an optimal balance between quality and cost-efficiency.
Grounding & classification
Source type: technical build writeup
18 fields verified against source quotes.
content generationdata extractionragproduct catalogfailure mode describedhuman review describednamed customerproduction runtime claimedsource backedtools describedworkflow describedecommercetechnical build writeupecommerce opsextract classify route