🎯 Fundamental Techniques
Master the essential techniques: Zero-Shot, Few-Shot, Chain of Thought, Role Prompting, and Humanization. The basic arsenal of every prompt engineer.
1. Zero-Shot Prompting
Direct instruction, no examples
Zero-Shot = ask for something without giving examples. Works for common tasks the model saw during training.
"Classify the sentiment as Positive, Negative, or Neutral:
'I love the product!'"
✅ When to use
Simple tasks, obvious format
⚠️ Limitation
May not capture a specific style
2. Few-Shot Prompting
2-5 examples to guide the pattern
Show input → output examples before your task. The model identifies the pattern and replicates it.
// Examples
Text: "João Silva is a director." → Name: João Silva | Title: Director
Text: "Maria Santos, CEO." → Name: Maria Santos | Title: CEO
// Your task
Text: "Pedro Lima, coordinator." → ?
💡 Tip
Use 3-5 representative examples and keep the format consistent
3. Chain of Thought (CoT)
"Think out loud"
Ask the model explain your reasoning step by step. Dramatically improves results in logic and math.
❌ No CoT
Model may jump to the wrong answer
✅ With CoT
Explicit reasoning, +40% accuracy
4. Role Prompting
"You are an expert in..."
Defining a role adjusts vocabulary, tone, depth, and perspective automatically.
👨💼 Consultant
"You are a consultant with 20 years of experience..."
👨🏫 Teacher
"You are a teacher explaining things to beginners..."
5. Text Humanization
Avoiding an “AI tone”
Techniques for making text sound more natural: contractions, opinions, structural variation, colloquial expressions.
❌ AI tone
"It is undoubtedly important to consider..."
✅ Humanized
"In my experience, it's well worth it..."
6. Iterate with Intention
Turn the first response into a better result
Choose a technique from this module, test the prompt, and create a second version by changing just one variable: examples, role, format, or tone.
Before You Submit
Define the expected result and the criteria for evaluating the response.
After the response
Record what improved, what got worse, and what the next isolated change will be.