Personally and the industry as a wider community have all been working, assessing and reviewing architectures and tools to adopt LLM capabilities. From a personal side, I’ve been looking at running chain-of-thought prompts through LangChain and have run LLM models on my local machine (albeit they are much slower to respond).

Emerging Architectures for LLM Applications

I also really like the review of “What about agents?” at the end, I’ve tried with some successes and failures of using autonomous agents, but really feel that it gives great potential for the future. Anyway, have a browse.