Prompt Engineering 2025: The Best Frameworks and Templates

Why prompt engineering matters more than ever

With the rise of powerful models such as GPT-5, Claude 3.5/4, and Gemini, the way we interact with generative AI has changed. In practice, one thing is clear: well-designed prompt templates deliver measurably more consistent results. That saves iterations, reduces rework, and makes AI outputs reviewable.

The problem: most users write their prompts off the cuff and get accordingly mediocre results. With structured templates you consistently get better outputs. The eight templates below cover roughly 80 % of typical SME tasks.

The four building blocks of a good prompt

A good prompt has four building blocks:

1. Role (profile): what role should the model play when responding? For example, a marketing expert, a developer, or a lawyer.

2. Directive: a clear instruction, for example “Write a summary” or “Generate Python code”.

3. Context: background information, target audience, or technical framing.

4. Output format: specifications for the shape of the response - lists, tables, bullet points, or a particular tone.

---

Template 1: Professional email

This template works for business emails of all kinds.

You are an experienced business communicator.
Write a professional email.

CONTEXT:
- Recipient: [name and role, e.g. "Mr. Mueller, Managing Director"]
- Relationship: [e.g. "long-standing customer",
  "new prospect", "internal colleague"]
- Occasion: [e.g. "complaint about delayed delivery"]
- Desired outcome: [e.g. "customer remains satisfied
  and keeps ordering"]

REQUIREMENTS:
- Tone: [professional/friendly/formal/empathetic]
- Length: maximum [X] words
- Language: professional English, formal address ("you"),
  suitable for a Swiss B2B context

Filled-in example:

You are an experienced business communicator.
Write a professional email.

CONTEXT:
- Recipient: Ms. Weber, head of purchasing at Mueller AG
- Relationship: customer for 3 years, smooth cooperation so far
- Occasion: her order #12345 is delayed by 5 business days
- Desired outcome: customer remains satisfied despite the delay

REQUIREMENTS:
- Tone: professional, empathetic, solution-oriented
- Length: maximum 150 words
- Language: professional English, formal address

ADDITIONAL INFORMATION:
- Reason for the delay: supply shortage at the supplier
- We offer 10% discount on the next order as compensation

---

Template 2: Document analysis

Ideal for contracts, reports, or long texts.

You are an experienced [analyst/lawyer/consultant].
Analyse the following document.

DOCUMENT:
[insert or upload the text here]

ANALYSE:
1. Summary in 3-5 sentences
2. The 5 most important key statements
3. Potential risks or issues
4. Open questions that should be clarified
5. Recommended next steps

FORMAT:
- Use headings for each section
- Use bullet points for lists
- Flag critical points with "WARNING:"
- Keep the language clear and understandable

---

Template 3: Meeting summary

Turn notes into structured minutes.

You are an experienced project manager.
Produce a structured meeting summary.

MEETING NOTES:
[paste your notes, transcript, or bullet points here]

PRODUCE:

## Meeting minutes
**Date:** [date]
**Participants:** [names]
**Duration:** [time]

## Topics discussed
[bullet points of the main topics]

## Decisions
[what was decided?]

## Action items
| Task | Owner | Deadline |
|------|-------|----------|
| ... | ... | ... |

## Open items
[what still needs to be clarified?]

## Next meeting
[date and proposed agenda]

---

Template 4: Code generation

For developers and technically minded users.

You are a senior software engineer. Write code for the
following task.

TASK:
[description of the desired functionality]

TECHNICAL REQUIREMENTS:
- Programming language: [e.g. Python, JavaScript, TypeScript]
- Frameworks/libraries: [e.g. React, FastAPI, pandas]
- Coding style: [e.g. PEP8, ESLint-compatible]

CODE REQUIREMENTS:
- Include docstrings/comments
- Implement error handling
- Write unit tests (optional)
- Consider edge cases

EXAMPLE INPUT/OUTPUT:
Input: [example]
Expected output: [example]

AFTER THE CODE:
- Briefly explain how the code works
- Mention possible improvements
- Point out security considerations

---

Template 5: Content creation (blog/social media)

For marketing teams and content creators.

You are an experienced content strategist.
Create content on the following topic.

TOPIC: [e.g. "Benefits of AI in customer service"]

TARGET AUDIENCE:
- Who: [e.g. "SME executives in Switzerland"]
- Knowledge level: [beginner/intermediate/expert]
- Pain points: [e.g. "too little time, high costs"]

CONTENT TYPE: [blog article / LinkedIn post / newsletter / tweet thread]

SPECIFICATIONS:
- Length: [e.g. "800 words" or "max. 280 characters"]
- Tone: [informative/entertaining/provocative/inspiring]
- Call to action: [e.g. "Register for the webinar"]
- SEO keywords: [list of relevant keywords]

STRUCTURE (for a blog post):
- Compelling headline with a number or question
- Hook in the first paragraph
- 3-5 subheadings
- Practical examples
- Conclusion with CTA

---

Template 6: Data analysis & Excel

For business analysts and Excel users.

You are a data analyst. Help me with the following analysis.

DATA DESCRIPTION:
- Data source: [e.g. "sales data Q3 2025"]
- Columns: [e.g. "date, product, revenue, region, customer"]
- Row count: [approximate number]

ANALYSIS GOAL:
[e.g. "Find the top 10 customers by revenue and identify trends"]

DESIRED OUTPUT:
- Excel formulas for: [e.g. "VLOOKUP, pivot table"]
- Visualisation: [e.g. "bar chart, trend line"]
- Interpretation of the results

SAMPLE DATA:
[insert 3-5 example rows]

RESPONSE FORMAT:
1. Step-by-step instructions
2. Ready-to-copy formulas
3. Interpretation of the results
4. Recommendations for further analyses

---

Template 7: Translation & localisation

For multilingual communication.

You are a professional translator with expertise in [subject area].

SOURCE TEXT:
[paste the original text here]

TRANSLATION TASK:
- Source language: [e.g. German]
- Target language: [e.g. English for the Swiss market]
- Target audience: [e.g. "B2B, financial services industry"]

REQUIREMENTS:
- Preserve the tone: [formal/informal/technical]
- Localise examples for the Swiss market
- Use the spelling conventions of the target locale
  (e.g. British vs. US English)
- Where uncertain: offer alternatives

GLOSSARY (if available):
| Source | Target |
|--------|--------|
| ... | ... |

OUTPUT:
1. Translation
2. Notes on tricky passages
3. Alternative wordings (where useful)

---

Template 8: Brainstorming & idea generation

For creative processes and strategy development.

You are a creative strategy consultant. Help me brainstorm.

STARTING POINT:
[describe the problem or challenge]

GOAL:
[what do you want to achieve?]

CONSTRAINTS:
- Budget: [if relevant]
- Timeframe: [if relevant]
- Resources: [what is available?]
- Restrictions: [what is off-limits?]

GENERATE:
1. 10 creative ideas (including unconventional ones)
2. Rate each idea on: effort (1-5) and impact (1-5)
3. Top 3 recommendations with reasoning
4. First implementation steps for the best idea

MINDSET:
- Think like a startup founder: what would be the fastest path?
- Think like a corporation: what would be the safest path?
- Think like a disruptor: what would surprise the industry?

---

Pro tips for better prompts

1. Use chain of thought

Ask the AI to think step by step:

Analyse this problem step by step:
1. First understand the situation
2. Identify the core challenge
3. Develop possible solutions
4. Evaluate the pros and cons
5. Give a recommendation

2. Supply examples (few-shot)

Here are examples of the desired style:

Example 1:
Input: "Customer is unhappy"
Output: "Thank you for your feedback. We understand
         your frustration and would like to..."

Example 2:
Input: "Technical issue"
Output: "We have forwarded your request to our technical team..."

Now process:
Input: [your current case]

3. Give negative instructions

AVOID:
- Empty phrases such as "I hope you are doing well"
- Overly long introductions
- Passive constructions
- Jargon where plain language exists

4. Refine iteratively

The result is good, but please:
- Cut it to half the length
- Make the tone more direct
- Add a concrete example
- Remove the last paragraph

---

Conclusion

Prompt engineering is the foundation for productive collaboration with AI. With these templates you have a toolbox that covers 80% of your daily AI interactions.

The most important tip: save your best prompts and refine them continuously. Over time you will build your own library of perfectly tuned templates.

In our training programmes we not only teach you these templates, but also the ability to develop your own prompts for your specific use cases.

---

Sources:

• Lakera: Prompt Engineering Guide (2025)
• Anthropic: Prompt Engineering Documentation (2025)
• OpenAI: Best Practices for Prompt Engineering (2025)