If the system assumes an LLM as the main loop then there's a floor of how cheap and performant it can be.
- If the system assumes an LLM as the main loop then there's a floor of how cheap and performant it can be.
- Whereas if it assumes normal compute sweetened with LLMs there's no floor or ceiling.
- And also if you assume LLM in the loop the only way to improve is model quality or tools.
- Whereas normal code can accrete functionality over time.