Ask HN: Best Use Cases for LLMs

I’m seeking advice on the scenarios where large language models (LLMs) shine and where they may not perform as well.

What types of problems have you successfully solved with LLMs? What are some common pitfalls or areas where they tend to underperform?