Skip to content


A hotel-searching application (Web app + API) which uses an LLMs and Vector Search to provide hotel summaries customized to the user preferences.

Key traits

  • LangChain and CassIO;
  • a Python API (FastAPI) and a React/Typescript client;
  • a partitioned vector store to optimize vector-based search of hotel reviews;
  • use of LLMs to prepare a user summary from user-provided preferences ...
  • which is then used to evaluate hotel summaries and provide information relevant to the user;
  • features also a real-time "post new review" flow: new review contribute immediately to the experience;
  • usage of (Astra DB) LLM response cache to save on latencies and costs with minimal boilerplate;
  • "Open in Gitpod" zero-hassle start mode available.

Visit the repository here.


User experience on the client:

Hotels demo, UI