simonjj/multi-llm-chat — reverse-engineered prompt

Build me a simple web app for chatting with different AI models from one place. I want a clean chat interface where I can choose which model to talk to, with Azure OpenAI as one option and any locally installed Ollama models showing up automatically. I also want to be able to paste links to a web page, a PDF, or a CSV so the selected model can use that content as extra context when answering.

Please make it easy to run locally with Docker, including the chat app, the backend service, the local model endpoint, and the vector database used to store ingested content so repeated lookups are faster. It should use environment variables for secrets and settings, and work with a GPU setup for local model inference. If possible, also keep it ready to deploy to Azure Container Apps later, since I may want a cloud version too.

Please wire up a basic but usable UI and the full end to end flow. Check current docs online if you need to.