@twostrawsPaul Hudson hey Paul, I'm enjoying your Swift AI Playbook. I'm curious though: in the section on managing context, your suggested approach is to take the most recent n messages from the conversation and discard the rest;
Most tutorials I've seen suggest using the LLM to summarise the conversation and then use the summarised transcript as the base context moving forward.
Do you plan to cover that option as well, or if not, is there a reason you wouldn't recommend that approach?