Guide/Breakdown of LLMs, Agents and Prompts

🤝 Discussion
1

Hi everybody! It's been brought to my attention that a nice multi-part guide and breakdown of LLMs, Agents and Prompts would be helpful to the community. So, I thought I'd start with the fundamentals and basics and gradually introducing new concepts and ideas while giving helpful tips & tricks, diagrams, sample code, interactive example apps with both Retool Agents and Python.... all leading up to an Agentic workflow to help improve/create/re-write your Prompts or SQL Statements.

I have some basic prompts I'll be using in the guide to help understand all the concepts/patterns/fundamentals I'll be covering, but I was wondering if anybody here might have a Prompt of their own that just isn't performing the way they would like or couldn't get to work at all that they would like to share so that I have some examples of real-use problems.

Related Topics:
Force Agent to Use a Tool by @Matthew_Carter

1 Like
2

@Jack_T is this what you were thinking?

Any tips, suggestions, comments are all welcome by all (especially the non-tech and ultra confused, but also you pros out there). Let me know if there's something missing you'd like covered specifically or if you think a different order of things might be easier to understand def let me know. As I said before, I do plan on creating examples so that you can see not just how different settings or different models respond with the same prompt but also how different models react to different strategies and styles. While teaching the ins and outs of Prompts the idea is to show that there is no one set path for Prompt Engineering, it depends on the model, the task you want it to do and the data you have or response type you need.... Prompt Engineering is more of a Toolbox, than a set of rules (feel free to pry off that o-ring with a flathead and toss away the Snap-On specific junk)

Planned Topic So Far (order pending)

  • Agents, Agentic, Models & LLMs (Preface)
    • quick overview/definition
    • comparison of each

  • Models (Intro)

    • overview of common providers and models
      • weaknesses/strengths
      • capabilities
      • built-in preferences (like format)
      • interactive examples

  • Retool Agent (Intro)

    • review UI
    • overview of models & providers (incl. custom providers)
    • review the parts of an Agent
    • quick overview of configuration settings

  • Agent Configuration (Intro)

    • define/explain each setting
    • show how each affect what the model does and/or how it responds

  • Prompts (Intro)

    • Parts of a Prompt
      • Input Data
      • Instruction
      • Context & Examples
      • Output Format

  • Prompt Engineering (Intro)

    • Define (It's more a process than a step or skill)
    • Benefits of Prompt Engineering

  • Prompts (Beginner)

    • Getting started
    • Common Pitfalls/Mistakes

  • Prompt Engineering (Beginner)

    • Overview of key elements
      • Prompt Format
      • Context and Examples
    • Word choice and sentence structure
    • Basic Prompt Layout/Organization (Headings)

  • Prompts (Intermediate)

    • Intro to Prompt Types
      • Direct (Zero-shot)
      • One, Few-Shot
      • Reflection
      • Chain of Thought (CoT)
      • Advanced Prompt Intro (combining types - Zero-Shot CoT)
    • Tips for writing better prompts

  • Prompt Engineering (Intermediate)

    • Structured Prompt Layouts
      • HTML, XML, JSON, Markdown
      • Combining Structure Types
      • Interactive Examples
    • Guardrails
    • Moderation
  • Prompts (Advanced)
    • Advanced Prompt Types
      • Multi-Shot
      • Chain of Thought (CoT)
      • Zero-Shot CoT
      • ReAct (Reason + Act)
  • Prompt Engineering (Advanced)
    • Dynamic Prompts
    • Memory(?)
    • Dynamic Context(?)
    • Context Engineering Intro

  • Advanced Concepts, Patterns and Strategies (cont)

    • Multi-Agent
    • Reasoning
    • Planning, Orchestrating, Synthesizing

  • All Tips & Tricks

    • big list

  • Other Sources for Tips/Tricks

    • ??

  • Other Sources by Model Provider

    • ??

Guides/Tutorials:

  • Single Agent to help with Prompts
  • Multi-Agent orchestration to improve a prompt (using multiple different models)
  • File Handling: PDF (loading, reading, performing RAG)
1 Like
3

Sounds great! Particularly interested in how to decide when multiagents are useful etc.

1 Like