Case Study: Sanwa Comtec achieves AI-powered harassment detection for on-prem chat with IBM watsonx.ai

Harassment determination by AI language model

Sanwa Comtec, an IT technology company, wanted to add AI-based harassment detection to its on-prem chat solution, Rocket.Chat, for customers that cannot use cloud services. The challenge was to identify potentially harassing conversations securely in a closed network using IBM’s AI and NLP capabilities.

Working with IBM, Sanwa Comtec built a custom model in IBM watsonx.ai and used IBM Watson NLP classification to train it on legally curated conversation data. The solution achieved more than 95% accuracy in harassment judgments in closed networks, and IBM continues to support Sanwa Comtec as it refines the model and prepares it for production.

Sanwa Comtec