llm supervisor v1 progress

💡In Progress

  • —

Setup

  • —

Input

0/1
  • Store transcript with timestamp in CSV/JSON

Logic + Processing

0/3
  • Add recent messages as context (simple memory)
  • Implement logic for LLM to "respond to a conversation"
  • Try out a second persona for contrast

UI

0/3
  • - Build a simple UI (Streamlit, notebook, or CLI)
  • Display conversation log
  • Add new input field (or simulate with logs)

Presentation and Reflection

0/2
  • - Write a reflective note (What worked? What next?)
  • - Share with your supervisors

☑️ Done

4/10
  • - Record a short video of the system in action
  • Do some research into how to use Whisper for realtime transcription
  • Create a basic prompt template with role/context
  • Write a 3-sentence scope for your prototype
  • Install Ollama or LM Studio locally
  • Run a test chat with a small model (e.g., Mistral or Phi)
  • Install Whisper or whisper.cpp
  • Choose your first LLM persona and write its prompt
  • Decide on text or voice input (or both)
  • Write a script to transcribe and save input