Chapter 1: DSPy Fundamentals | DSPy: The Comprehensive Guide

What You'll Learn

By the end of this chapter, you will:

✅ Understand what DSPy is and why it matters

✅ Grasp the paradigm shift from prompting to programming

✅ Install and configure DSPy in your development environment

✅ Write and run your first DSPy program

✅ Configure and work with different language models

✅ Build simple question-answering applications

Chapter Overview

This chapter covers the essential foundations you need to start building with DSPy:

📖

What is DSPy?

Learn what DSPy is, why it was created, and how it differs from traditional prompt engineering approaches.

🔄

Programming vs. Prompting

Understand the fundamental paradigm shift from manual prompt engineering to programmatic LM pipelines.

💻

Your First DSPy Program

Write and run a complete DSPy application from scratch.

🤖

Language Models

Learn how to configure and work with different LM providers (OpenAI, Anthropic, local models).

🏋️

Exercises

Practice what you've learned with hands-on exercises.

What Makes DSPy Different?

Before diving in, here's a quick preview of what makes DSPy special:

❌ Traditional Prompting

# Manual prompt engineering
prompt = """
You are a helpful assistant. Answer the question clearly.

Question: What is the capital of France?
Answer:
"""

response = openai.chat.completions.create(
    model="gpt-4",
    messages=[{"role": "user", "content": prompt}]
)

⚠️ Brittle and hard to maintain ⚠️ Doesn't compose well ⚠️ Manual tuning required ⚠️ No systematic optimization

✅ DSPy Approach

import dspy

# Define the task signature
class QuestionAnswer(dspy.Signature):
    """Answer questions clearly."""
    question: str = dspy.InputField()
    answer: str = dspy.OutputField()

# Use it with automatic prompting
qa = dspy.Predict(QuestionAnswer)
response = qa(question="What is the capital of France?")

✅ Clean, modular code ✅ Easy to compose and reuse ✅ Automatically optimizable ✅ Systematic improvement

Prerequisites

Before starting this chapter, ensure you have:

✅ Python 3.9+ installed

✅ Basic Python knowledge (functions, classes, imports)

✅ Virtual environment set up (from setup instructions)

✅ API key for at least one LM provider

✅ Text editor or IDE ready to use

💡

Need help with prerequisites? Review Chapter 0: Prerequisites

Learning Approach

This chapter uses a hands-on approach:

📚

Concepts

Clear explanations of core ideas

💻

Examples

Working code you can run

🏋️

Practice

Exercises to reinforce learning

🔬

Experimentation

Encouragement to modify and explore

💡

Pro Tip: Don't just read—run every example and complete the exercises!

Estimated Time

📖 Reading 1-1.5 hours

💻 Running examples 1 hour

🏋️ Exercises 1-1.5 hours

⏱️ Total 3-4 hours

Feel free to spread this over multiple sessions!

Key Takeaways (Preview)

By the end of this chapter, you'll understand:

DSPy is a framework for programming (not prompting) LM-based applications

Signatures define tasks declaratively using input/output specifications

Modules are composable building blocks that can be optimized automatically

LMs are configurable and DSPy works with multiple providers

Programming > Prompting for building robust, maintainable applications

🚀

Let's Begin!

Ready to learn DSPy? Start with understanding what DSPy is and why it matters.

Remember: Learning is a journey. Take your time, experiment freely, and don't hesitate to ask questions!

Continue to What is DSPy?

DSPy Fundamentals

🚀 Welcome to DSPy