- [ ] Do more research on selecting which models to use - [ ] speech to text & text to speech - [ ] Good user experience with fallback answers and loading states
