Examples

1. 1.

   Gmail's Smart Labels use text classification to route incoming emails to Primary, Social, Promotions, and Updates tabs - processing 1.8 billion active users' inboxes with sub-millisecond latency per message.

2. 2.

   Zendesk's intent classification model routes 50+ million monthly support tickets to the correct team (billing, technical, account, complaint) with 94% accuracy, reducing misrouting costs by $8M annually for large enterprise customers.

3. 3.

   Hugging Face's zero-shot text classification pipeline (using NLI-based models like DeBERTa) is used by media companies to categorise news articles into topic taxonomies without training data, achieving 85% accuracy on 50+ categories.