It is excellent to learn prompt engineering, RAG and LangChain, so that the application of LLMs can be much more than chatbot.

The hands-on is manageable, yet allow learners to experience the actual flow of using the tools.

veri nice and intuitive

Great animations and notebooks, but I got some errors during imports and installation, they should be reworked. Also, the last notebook states that if you want to interact with some private documents, RAG is a good choice. For my understanding, pieces of private documents are sent through the internet during RAG, since they're retrieved and added to prompt in-context, that's why I find this notebook a bit misleading.

Great Course to Learn the AI agent and RAG. I liked the Summary of what you'll learn and recap in each video. The exams where in good level so you had to clearly understand the concept to be able to get a good mark on them. I felt like instructor spoke quite fast though, so I had to reply each video to keep up with the materials.

A glossary would be useful.

I think there are some issues with the tests - a number of times I feel the answers are either incorrect or the questions are poorly worded so as to be ambiguous. For example: "In agents, a language model is used as a reasoning engine to determine which of the following actions?" - LangChain agents use only Python code for building applications - Language model - Task manager - Data loader From the videos/summary, agents are clearly communicated as Task manager (see below). Additionally, the wording 'In agents' feels like it is referring to a field or domain of specialization which was never introduced; the concept of an agent as an object was taught. Furthermore, the grammar of the question and answers don't align - by asking for 'actions' you are asking for a verb, but all of the answers are nouns. " - Agents in LangChain are dynamic systems where a language model determines and sequences actions, such as predefined chains. - Agents integrate with tools such as search engines, databases, and websites to fulfill user requests. " Furthermore, an LLM by definition cannot be referred to as a reasoning engine as it is probabilistic by its very nature. The ability to reason (or produce reasoned responses) is not the same as a reasoning engine, which is something like a calculator that is deterministic and based on fundamental axioms.