Method
LLM classification / extraction
Using an LLM for classification, tagging and attribute extraction.
34 cases from 24 companies apply LLM classification / extraction. 6 were checked in an online A/B test; 11 report results in production.
Typical tasks: Evaluate models, Match entities, Build an assistant over internal docs, Catch fraud and abuse and 11 more. Companies: DoorDash, Ramp, Databricks, Gusto, Yandex, Avito and 18 more. Often combined with: RAG, Human in the loop, LLM-as-a-judge, LLM labeling, Few-shot prompting and 68 more.





