Prompting Fundamentals

The art of telling a language model what you want — surprisingly similar to human communication.

25 min read Beginner 8 lessons April 26, 2026

What you'll learn

  • 5 Prompt Frameworks compared: RACE, CRISPE, APE, RISEN, CRAFT
  • 5 proven techniques for better results
  • Common mistakes and how to avoid them
  • Practical examples for everyday use
0/3

Introduction

The difference between 'Write something about marketing' and a precise, structured prompt is like the difference between a vague sketch and a detailed blueprint. AI systems deliver exactly what you ask for – nothing more, nothing less. In this guide, you'll learn how to achieve consistently good results through systematic communication.

Why is this important?

  • AI tools are only as good as the instructions they receive
  • Prompting is the 'programming language' of the AI era
  • Small changes in the prompt can make big differences in the result

After this guide you'll be able to:

  1. Explain the 3-layer model of AI communication
  2. Structure prompts using the RACE framework
  3. Apply advanced techniques like Chain-of-Thought
  4. Systematically evaluate and improve results
  5. Recognize and avoid the most common mistakes

Interactive: Three Perspectives on Prompting

A prompt can be understood in different ways. Switch between three analogies and discover how each perspective illuminates different aspects of prompting.

What is a Tensor?

A tensor is the fundamental data structure in modern AI systems. Depending on the perspective, tensors can be understood in very different ways. Switch between three analogies to experience different perspectives.

Prompting as a Cooking Recipe

A prompt is like a cooking recipe: you give the model a list of ingredients (context), cooking instructions (task), and a description of the desired dish (format). The more precise your recipe, the more predictable the dish. A vague "cook something nice" produces surprises — a detailed recipe delivers what you want.

INGREDIENTS (Context):
- 1x Product description (500 words)
- 1x Target audience: students
PREPARATION (Task):
Summarize the text.
PLATING (Format):
3 bullet points, max. 50 words each.
RESULT: Structured summary
Concrete Example

Instead of "Summarize" (vague), you write: "You are an editor. Summarize the following product text for students. Use three bullet points with a maximum of 50 words each. Start each point with a verb." — like a recipe with exact measurements.

Strength of this Analogy

Immediately makes clear why structure and precision matter — everyone knows the difference between "cook something" and a proper recipe.

Limitation of this Analogy

Cooking recipes are deterministic (same ingredients = same result). Prompts produce slightly different outputs with each run — the model is probabilistic, not a bread machine.

Tensor Dimensions Compared

RankMathematicsCooking RecipeAI Example
Active Analogy:1 / 3Exploring all perspectives helps with understanding

Foundations

What is a Prompt?

AnalogyScientific
A prompt is like a work assignment for a very capable but context-free employee. The more precise the instruction, the better the result. Imagine you had a brilliant assistant who knows everything – but nothing about your current situation. The prompt is your chance to tell them exactly what you need.

Analogy:

A prompt is like a work assignment for a very capable but context-free employee. The more precise the instruction, the better the result. Imagine you had a brilliant assistant who knows everything – but nothing about your current situation. The prompt is your chance to tell them exactly what you need.

Scientific:

A prompt is the text input that serves as instruction to a language model. It determines which patterns from the training knowledge are activated and in what form the output occurs. The quality of the prompt has a direct impact on the relevance, accuracy, and usefulness of the generated response.

What is Prompt Engineering?

AnalogyScientific
Prompt Engineering is the art of asking the right questions – like an experienced interviewer who elicits the best answers through skillful questions. It's not about what the AI can do, but how you tell it what to do.

Analogy:

Prompt Engineering is the art of asking the right questions – like an experienced interviewer who elicits the best answers through skillful questions. It's not about what the AI can do, but how you tell it what to do.

Scientific:

Prompt Engineering is the systematic development and optimization of prompts to achieve desired outputs from language models. It encompasses techniques for structuring, contextualizing, and iteratively refining prompts.

The 3-Layer Model

When you communicate with an AI, a dialogue takes place on three levels. You control only one of them – but it's your most powerful lever.

3User Prompt

Your specific input for the current task

Full control
2System Prompt

Hidden instructions from the provider that shape behavior

Indirect (provider choice)
1Foundation

The model's training – billions of texts it has 'read'

No influence

System prompts are normally invisible to users. They are set by the provider and define the 'personality' of the assistant. This explains why ChatGPT 'sounds' different from Claude – both have different system prompts.

"You are a helpful, honest assistant. Your goal is to fully understand the user's intent..."

Caution: AI mirrors your tone

  1. Confirmation Tendency: 'Wasn't that a good idea?' often leads to agreeing answers – regardless of whether the idea is good.
  2. Priming Effect: The first sentence of your prompt colors the entire response.
  3. Politeness Overhead: 'Would you be so kind...' is unnecessary – clear, direct commands work better.

Actively request counterarguments: 'Play devil's advocate.' or 'What are the weaknesses of this idea?'

Checkpoint

  • You understand the 3-layer model
  • You know the psychological traps in prompting

Prompt Frameworks Compared

There are various proven frameworks for structured prompting. Each has its strengths – choose the right one for your task.

RACECRISPEAPERISENCRAFT

RACE is the most versatile framework for everyday use. It covers the key aspects: who responds, what should be done, for whom, and in what format.

Best for: General tasks when you need a clear structure

R

Role

Define who responds – what expertise, perspective, or personality should the AI adopt?

A

Action

What exactly should be done? Use precise verbs like 'analyze', 'create', 'compare'.

C

Context

Who is the result for? What prior knowledge exists? What's the occasion?

E

Execute

What format should the answer be in? Length, structure, style?

Role

The role is often the most powerful and economical lever. With one or two words, you narrow down the vast possibility space of the AI to a focused area.

Without Role
Explain blockchain.

Technical, dry, without audience focus

With Role
As a financial journalist writing for laypeople: Explain blockchain.

Understandable, with everyday examples, appropriate depth

The Method-Acting Trap

AI can take roles too seriously. Studies show: With strong identities (e.g., 'As a conspiracy theorist...') factual accuracy can suffer because the AI wants to stay 'in character'.

Use role for perspective, style, and structure

No role for pure fact questions or logic tasks

Action

The Action defines the verb – what exactly should the AI DO? The more precise the action instruction, the better the result.

Vague
Do something about marketing.
Precise
Create 5 headlines for a fitness app that motivate daily exercise.

Strong Action Verbs

For analysis:Analyze, Compare, Evaluate, Identify
For creation:Create, Generate, Write, Design
For structuring:Summarize, Extract, List, Categorize
For transformation:Explain, Translate, Simplify, Adapt

Context

Context answers: For whom? Why? With what prior knowledge? Without context, the AI delivers generic output – with context, it delivers relevance.

Without Context
Write about AI.
With Context
Article for the Financial Times. Target audience: Skeptical mid-market managers. Focus: Practical applications and ROI.

Execute

Execute transforms good content into usable results. How should the answer be structured? What style, length, format?

Without Format
Give me the answer.
With Format
Structure as: Executive Summary (3 sentences), then 3 main points with one practical example each. Max. 500 words, factual tone.

Complete example

[ROLE]

You are an experienced strategy consultant, specialized in digital transformation in traditional industries.

[ACTION]

Identify and explain the three most important AI application opportunities for medium-sized machinery manufacturers.

[CONTEXT]

Presentation for a conservative board. Company with 500 employees, currently no AI usage, digitalization pressure from competitors.

[EXECUTE]

Per application: Concise headline, three benefit bullets, one concrete practical example. Factual-optimistic tone, no jargon.

You can copy and try the full prompt here: View Prompt

Which Framework Fits You?

Each framework has its strengths. This overview helps you decide:

Framework
Strength
Best for
Complexity
RACE
Versatile & easy to learn
Everyday tasks of all kinds
Easy
CRISPE
Maximum control over style
Creative & marketing content
Advanced
APE
Audience-focused
Content for specific readers
Easy
RISEN
Structured processes
Technical & step-by-step tasks
Medium
CRAFT
Complete text production
Business documents & formal texts
Medium

Our Tip

Start with RACE – it's the simplest framework and covers most use cases. Try other frameworks when you have special requirements.

Checkpoint

  • You know 5 frameworks: RACE, CRISPE, APE, RISEN, CRAFT
  • You can choose the right framework for your use case

Advanced Techniques

RACE structures the prompt – but sometimes the AI needs help with thinking. The following techniques force the model to work more structuredly and deliver better results.

Chain-of-ThoughtFew-ShotSelf-CritiqueSelf-ConsistencyTag Annotation

Chain-of-Thought (Step-by-Step)

Chain-of-Thought is the most important technique for complex tasks. You get the AI to make its thinking visible – this prevents careless errors and significantly improves the quality of answers.

Chain-of-Thought (Step-by-Step)

AnalogyScientific
Like a student who writes down the calculation steps for a math problem instead of just the answer. Writing down the intermediate steps prevents careless errors.

Analogy:

Like a student who writes down the calculation steps for a math problem instead of just the answer. Writing down the intermediate steps prevents careless errors.

Scientific:

Chain-of-Thought (CoT) is a prompting technique that instructs the model to explicitly verbalize its thinking process before arriving at the final answer. This activates more structured processing patterns.

When to use?

  • Mathematical calculations and logical problems
  • Multi-step decisions with dependencies
  • Analysis of complex situations with many factors

Vague vs. Precise

Standard
What is 17 × 24?
With CoT
Calculate step by step: What is 17 × 24?

Trigger Phrases

  • "Think step by step."
  • "Explain your reasoning."
  • "Work through the problem systematically."

Complete Example

You are a financial advisor. Analyze step by step: 1. What factors are relevant? 2. How are they connected? 3. What is the conclusion? Question: Should a 35-year-old with $50,000 in savings invest in ETFs or individual stocks?
View Prompt

Least-to-Most Prompting

For very complex problems, you can ask the AI to first create a plan: 'Break down this task into individual steps. Don't solve anything yet.' Then: 'Now solve each step one by one.' This prevents the model from getting tangled in its own train of thought.

Step-Back Prompting

For knowledge questions with hallucination risk, a step back helps: 'What general principles are relevant for this problem?' Then: 'Use these principles for the following question...' The model 'remembers' the basics before it answers.

The Creativity Dial: Temperature

  • Low (0.1-0.3): Consistent, focused – ideal for analysis and facts
  • Medium (0.5-0.7): Balanced – suitable for most tasks
  • High (0.8-1.0): Creative, surprising – ideal for brainstorming

When to use which technique?

Logical or mathematical problemChain-of-Thought
Format hard to describeFew-Shot
Quality control neededSelf-Critique
Critical decisionSelf-Consistency
Complex, multi-part taskTag Annotation

Checkpoint

  • You know 5 techniques: Chain-of-Thought, Few-Shot, Self-Critique, Self-Consistency, Tag-Annotation
  • You know when to use which technique

Practical Examples

The best way to learn prompting is through practice. Here we show you typical use cases and how to apply the techniques you've learned.

Application Areas Overview

Data Analysis

Interpret sales figures, identify trends, summarize reports

Chain-of-Thought + clear metrics

Creative Writing

Product descriptions, social media posts, storytelling

Few-Shot + style guidelines

Learning & Understanding

Get complex concepts explained, exam preparation

RACE + audience context

Programming

Code explanations, debugging, refactoring

Structured tags + concrete code

Business Communication

Drafting emails, structuring presentations

Role specification + tone definition

Translation & Adaptation

Translate texts, cultural adaptation, simplification

Context + target audience definition

Example 1: Email Prioritization

You're overwhelmed with 50 unread emails. You need an assistant to help filter out the most important ones.

Structured Prompt (RACE Format)

[ROLE]You are my personal assistant with experience in time management.
[ACTION]Analyze the following email subject lines and categorize them by urgency.
[CONTEXT]I'm a project manager with a meeting in 2 hours. Everything related to Project 'Alpha' has highest priority.
[EXECUTE]Give me a numbered list: 1. IMMEDIATE (before the meeting), 2. TODAY, 3. CAN WAIT. Maximum 3 emails per category.

Why does this prompt work well?

  • Clear role gives the model a perspective for prioritization
  • The context (Project Alpha, 2h until meeting) enables relevant decisions
  • The Execute format prevents all 50 emails from being listed individually

Example 2: Product Descriptions

You need engaging product texts for an online shop. Instead of describing the format at length, you simply show examples.

Few-Shot Prompt

Write product descriptions in the following style:

Product:Wireless Bluetooth Headphones→Immerse yourself in crystal-clear sound – anywhere. Up to 30h battery, memory foam cushions for marathons, not just podcasts.
Product:Stainless Steel Thermos Flask 500ml→Your coffee stays hot until the third meeting. Double-walled stainless steel, leak-proof, fits in any backpack.
Product:Ergonomic Gaming Mouse→?

Why does this prompt work well?

  • The examples define length, tone, and structure without explicit rules
  • The style is recognized by the model: emotional, feature-focused, with a wink
  • No long explanations needed – the model learns from the pattern

Troubleshooting

The 4 Classic Pitfalls

Too vague

AI guesses what might be meant

Apply RACE framework

Too much at once

Contradictory or mediocre results

Split task into multiple prompts

Anthropomorphizing

Politeness phrases with no effect

Formulate clear, direct commands

Negation

'Don't' is ignored

Formulate positively: what should be included

Why Positive Framing Is More Robust

Older language models had real difficulties with negation — 'don't' was often ignored. Modern models like GPT-4o and Claude handle negation much more reliably. Still, positive framing remains the more robust strategy: it gives the model a clear direction rather than just boundaries.

Negative (works poorly)
Write about luxury watches, but DON'T mention the price.
Positive (works better)
Write about luxury watches. Focus on craftsmanship, materials, and status.

Quality Checklist

Check every AI response

  • Relevance: Does the answer exactly fulfill the task?
  • Completeness: Were all instructions considered?
  • Correctness: Are facts verifiable? (Always validate externally!)
  • Format: Does the output match the format specifications?
  • Tone: Does the answer hit the defined role/perspective?

Summary

Key Takeaways

  1. Quality Input = Quality Output – Precise prompts deliver precise results
  2. RACE as checklist – Role, Action, Context, Execute structure every prompt
  3. Techniques for complex tasks – Chain-of-Thought, Few-Shot, Self-Critique expand your repertoire
  4. Iteration is normal – The first prompt is rarely perfect
  5. Positive over negative – Describe what you want, not what you don't want
  6. Always verify facts – AI can be convincingly wrong

Frameworks

RACE
R Role
A Action
C Context
E Execute
CRISPE
C Capacity (Role)
R Request
I Insight
S Statement
P Personality
E Experiment
APE
A Audience
P Purpose
E Execution
RISEN
R Role
I Input
S Steps
E Expectation
N Novelty
CRAFT
C Context
R Role
A Action
F Format
T Tone

Techniques

Chain-of-Thought"Think step by step."
Few-Shot[Show examples]
Self-Critique"Review your answer."
Self-Consistency"Solve in 3 different ways."

Quiz

Answer the following questions to check your understanding. With at least 70% correct answers, you'll earn an achievement!

Test Your Knowledge

Question 1 / 13
Not completed

Which RACE component defines WHO responds?

Select one answer

1. Which RACE component defines WHO responds?

  • ☐ A) Action
  • ☐ B) Role
  • ☐ C) Context
  • ☐ D) Execute

2. What is Self-Consistency particularly useful for?

  • ☐ A) Writing creative texts
  • ☐ B) Summarizing long texts
  • ☐ C) Validating critical decisions through multiple solution paths
  • ☐ D) Describing images

3. What's the problem with the prompt 'Don't mention prices'?

  • ☐ A) It's too short
  • ☐ B) The action is missing
  • ☐ C) Negations are often ignored – better to formulate positively
  • ☐ D) The role is missing

4. Chain-of-Thought is particularly useful for:

  • ☐ A) Logical and mathematical problems
  • ☐ B) Creative texts
  • ☐ C) Simple translations
  • ☐ D) Formatting lists

5. What does 'method-acting trap' mean for role assignment?

  • ☐ A) The AI can't play roles
  • ☐ B) The AI forgets the assigned role
  • ☐ C) The AI sacrifices facts to stay in character
  • ☐ D) The AI spontaneously switches roles

6. Which statement about the 3-layer model is correct?

  • ☐ A) You control all three layers
  • ☐ B) System prompts are always visible
  • ☐ C) The user prompt is the only layer with full control
  • ☐ D) The foundation can be changed through good prompts

7. What is Few-Shot Prompting?

  • ☐ A) A short prompt with few words
  • ☐ B) A prompt with examples showing the desired format
  • ☐ C) A prompt that only works once
  • ☐ D) A prompt without context

8. What should be your FIRST step when the AI response doesn't meet your expectations?

  • ☐ A) Use a different AI model
  • ☐ B) Analyze the prompt and identify the cause
  • ☐ C) Add more context
  • ☐ D) Completely reformulate the request

9. What is Tag Annotation particularly useful for?

  • ☐ A) Translating texts
  • ☐ B) Clearly structuring different prompt parts
  • ☐ C) Making the AI be quiet
  • ☐ D) Describing images

10. Which prompting best practice is CORRECT?

  • ☐ A) Always be as polite as possible ('Would you be so kind...')
  • ☐ B) Combine all available techniques in every prompt
  • ☐ C) Give clear, direct instructions and be specific
  • ☐ D) Keep prompts as short as possible, even if context is missing

11. You receive a long, generic answer with no specific connection to your topic. What's probably missing from your prompt?

  • ☐ A) Role – you didn't define expertise
  • ☐ B) Context – the AI lacks background information
  • ☐ C) Execute – you didn't specify a format
  • ☐ D) Action – your verb was too unspecific

12. You need 5 product names for a new fitness app. The AI only delivers 3 and then goes off-topic. What do you do?

  • ☐ A) Use a different model
  • ☐ B) Repeat the prompt and hope
  • ☐ C) Make the Execute element more precise: 'Exactly 5 names, numbered, without further explanation'
  • ☐ D) Add more context about fitness

13. This prompt delivers inconsistent results: 'Write something cool about sustainability.' Which RACE elements are missing?

  • ☐ A) Only Role is missing
  • ☐ B) Only Execute is missing
  • ☐ C) Role, Context, and Execute are missing
  • ☐ D) All four elements are present
Answer Key: 1) B · 2) C · 3) C · 4) A · 5) C · 6) C · 7) B · 8) B · 9) B · 10) C · 11) B · 12) C · 13) C