In this article, a developer details the creation of a Quarkus extension for Google Vertex AI, which enables the development of a chatbot. The process begins by setting up a Quarkus project with relevant extensions for REST and Vertex AI. This extension allows injection of a VertexAI object to access Google’s generative AI model, Gemini. The chatbot is implemented via a REST endpoint at /chat, where a system message instructs the AI on its behavioral guidelines. Utilizing cURL, the chatbot can receive messages and respond with humorous, relevant content. The initial implementation lacks memory; however, an updated version introduces session memory, allowing the chatbot to remember previous interactions. While this example uses a single memory for demonstration, real applications should implement individual session handling for each user. The article encourages further exploration of Google Vertex AI’s capabilities and provides a link to the project on GitHub.
