1、Agents,Large Language Models,and Smart AppsAI Infrastructure ReportTable of contentsAgents,Large Language Models,and the New Wave of Smart Apps03The New Stack for Intelligent Apps08Vector Databases35Advanced Design of AI-Driven Apps48The Future and Where It All Ends Up52Building AI-Driven Apps13The

2、Fine Art of Prompt Engineering14Agent Frameworks28LangChain29The LLMs Themselves41Open Source Models43Challenges with Running Open Source Models in Production44LlamaIndex30Haystack31Semantic Kernel32The Future of Frameworks34From Prompting to Tuning213Agents,Large Language Models,and Smart Apps01Age

3、nts,Large Language Models,and the New Wave of Smart AppsWhen you hear the word agent,you might think of 007 or Jason Bourne.They can fight crime with one hand and down a Martini with the other,and always look stylish doing it.But since the release of ChatGPT,weve seen an explosion of a new kind of a

4、gent.AI agents are intelligent programs that can interact autonomously or semi-autonomously with their environment.Actually,the definition of agents is still evolving at the moment.Traditionally,an agent is defined as that tries to achieve its goals in the digital or physical world,or both.Its got s

5、ensors that“see,”“hear,”and“sense”its environment.It has“actuators,”a fancy word for the tools it uses to interact with the world,whether thats an LLM using an API the way we use our hands and fingers,a robotic gripper picking up trash,or a self-driving car sensing the environment with LIDAR.But Lar

6、ge language models(LLMs)like ChatGPT and GPT-4,based on the ultra-popular,changed what is possible with Agent capabilities.For the first time they give us little“brains”that are capable of performing a wide range of tasks,from planning and reasoning to answering questions and making decisions,which