How did we go about building it
...

Step 1
...

9th Dec:
- Created the base streamlit chat app, and made calls to OpenAI GPT models.
- Took care of chat history and logging
- Used OpenAI's GPT 3.5 & 4 models to get the base code for individual functionality.
  - But there were errors - it could not code the chat history in streamlit. It did not understand the st.session construct
  - And it could not write code for loguru - different levels to different channels or files. For that I had to go back and research the web and find answers
10th Dec:
- Found something very interesting in this article
  - ollama - to run models locally
  - zephyr - a powerful open source model
- I installed ollama and played with it