Caikit was deployed to serve the Llama 2 model, TGI provided text generation, and Gradio and LangChain were used for the user interface and tool integration. The Llama 2 model was served using Caikit. Redis database was also deployed and the documents were embedded into its vector store.