Last Sunday, we took Gemmaiku, my fine-tuned version of Gemma, and started turning it into a ChatGPT-style product. This Sunday, we go one layer deeper. We’ll build the part that makes the chatbot work using Python + FastAPI, connect it to Ollama, and serve Gemma/Gemmaiku through our own API. Less UI. More getting our hands dirty with what’s happening underneath. Pick up your laptop. Let’s build.