Prompt engineering best practices for LLMs on Amazon Bedrock
summary
Practical prompt rules for Amazon Bedrock with Anthropic Claude, showing concrete outcomes: one‑shot and few‑shot examples, Chain‑of‑Thought improves arithmetic, and role prompting raises puzzle accuracy; prompts use Claude 3 messages API and structured XML text tags to force JSON outputs. Bedrock supports 200k token context and automates Retrieval‑Augmented Generation via Knowledge Bases for Amazon Bedrock, while function‑calling agents let the model request tool calls and return reliable answers.
AI-written summary · about
chapters (19)
- 0:00Introduction to prompt engineering
- 0:50The art of prompting
- 4:05Prompting techniques
- 8:05Anthropic Claude best practices
- 10:05Clear and direct instruction
- 12:26Role prompting
- 14:16Few-shot and chain of thought
- 16:01Structured prompt engineering
- 17:16Context and prompt chaining
- 18:53Structured templates
- 21:09System prompts
- 24:46Adding science to the art
- 26:05Evaluation and hallucinations
- 27:47Retrieval Augmented Generation
- 29:39Knowledge bases in Bedrock
- 32:21Agents and function calling
- 35:19Implementing function calls
- 37:13Guardrails and security
- 38:54Resources and closing
related talks
29mAnthropic's Claude in Amazon Bedrock
31mLamRAG: From Data to Constructive Insights using Amazon Bedrock
19mUnlocking the Full Potential of LLMs: Advanced Prompt Engineering Techniques
01h 39mMastering LLMs: From Prompt Engineering to Agentic AI
59mTutorial: How to Win Friends & Influence LLMs (with Prompt Engineering)
01h 35mBuilding with Anthropic Claude: Prompt Workshop
26mMastering Large Language Models: Prompt Engineering for Precision & Performance
48mStop Fighting Your AI Engineering Prompts That Actually Work