Easyway Case Study

Luxury hotels need to handle 200+ messages a day while maintaining 4- and 5-star service, from website visits through checkout. Easyway is a guest relationship management (GRM) platform that connects guests and hotels via WhatsApp, SMS, Telegram, and automates check-in, check-out, and everyday inquiries. The key is making guests feel like they're talking to a person, not a chatbot.

The existing model failed with even slight variations in how questions were phrased, and it had a problem of responding with excessive confidence even when it didn't know the answer. AI21 Labs helped by breaking the process into steps and translating it into code, and by identifying data inconsistencies, and Easyway reorganized the model with dozens of custom prompts and training data.

There are two core applications.

• Intent detection: Classifies intent based on 15 frequently asked questions, and doesn't respond at all when unsure. This is to avoid escalating situations by giving a wrong answer to an angry customer.
• Automatic translation: Provides two-way translation across 100+ languages, improved to better handle inputs with rough grammar. The same technology used in Wordtune was applied to this improvement.

Going forward, Easyway plans to add a chat widget to websites to automatically handle FAQs before and after booking, and to send messages timed to events like booking, arrival, and departure. One European customer receives thousands of emails due to 150,000 visitors a month, and Easyway aims to reduce the volume of manual responses by 10-30%.