It's so funny how LLMs are pretty much useless for most real-life use cases without tool calling or RAG, and so tools will be like "thinking... executing function xyz..." Congrats, you just made a web app of ten microservices wearing a trench coat.