Case Study: Guesty launches a RAG-based chatbot in three weeks with Qwak

How Guesty uses Qwak for hyper speed ML model delivery

Guesty, a property management platform for short-term and vacation rentals, wanted to deploy its first RAG-based chatbot to improve data science operations and accelerate model delivery. The team needed a low-latency vector database, a way to scale from proof of concept to production, and a solution that would let data scientists lead implementation with minimal engineering support, so it worked with Qwak and its Vector DB offering.

Qwak helped Guesty build and launch the chatbot in about three weeks, using Google BigQuery, OpenAI embeddings, Qwak’s managed vector database, and ChatGPT 3.5 to retrieve relevant context and generate responses. The result was a one-click path from POC to production, zero engineering dependency, improved operational efficiency and SLA performance, and a rise in chatbot user engagement from 5.46% to 15.78%, with stronger guest satisfaction and plans to extend the architecture to more models.

Guesty

Elad Silvas

Data Science Team Lead