In a RAG system, the prompt is where everything comes together: the user's question, the retrieved documents, and your instructions all end up in one block of text that the model has to make sense of. A clear, well-structured prompt is often the cheapest way to get better answers, long before you reach for fine-tuning or a more complex pipeline.
The lessons start with a beginner-friendly course on what prompt engineering is and how to apply it, then zoom in on two core techniques: zero-shot and few-shot prompting, where you control the output by giving examples, and chain-of-thought prompting, where you ask the model to reason step by step. Next, Anthropic's applied AI team builds a real prompt live, showing how structure and XML tags keep long prompts organized, and a final roundtable with Anthropic's prompt engineers explains how to think about the craft.
Watch the lessons in order. The last video is a long conversation rather than a tutorial, so feel free to watch just the first part for now and come back to it later.
A beginner-friendly freeCodeCamp course on what prompt engineering is and why it matters, with a quick tour of language models, best practices for writing prompts, zero-shot and few-shot prompting, and AI hallucinations.
Video 42minHow the number of examples in a prompt changes the answer: giving none, one, or a few, and when it is worth adding examples to control the format and quality of the output.
Video 16minWhy asking a model to work through a problem step by step gives better answers on multi-step tasks, and how to write chain-of-thought prompts for practical use cases.
Video 24minAnthropic's applied AI team builds a production-style prompt live in the console: how to structure it, organize information with XML tags, and iterate on the result.
Video 25minA roundtable with four Anthropic prompt engineering experts on what makes a good prompt engineer, how to refine prompts, and where the field is heading. A long, optional conversation to watch in parts.
Video 1h 16min