AI Prompt Templates That Actually Work: 12 Scenarios I Use Weekly

I've seen a lot of "ultimate prompt template" articles. Most of them read like someone copied a framework from a textbook and filled in the blanks with generic examples. I rarely found myself actually using those templates.

So here's a different approach: 12 prompt patterns I genuinely use in my own work, across writing, coding, analysis, email, and more. These aren't theoretical — they're the evolved versions of prompts that survived real-world testing.

The One Framework Behind All of Them

Before I get into specific templates, there's one underlying structure that makes all of them work. I didn't invent this — it's the BROKE framework, and it's become the backbone of how I write any prompt that needs to produce real output.

B - Background: What's the context? What's the situation?
R - Role: What persona should the AI adopt?
O - Objective: What exactly do I need done?
K - Key Result: What does "done" look like? Format, length, quality?
E - Example: What does a good output look like?

Not every prompt needs all five elements spelled out explicitly. But if a prompt isn't working, adding whichever element is missing usually fixes it.

Here's what this looks like in practice:

[B] I'm a product manager at a SaaS startup, and I need to present
our quarterly plan to the leadership team.

[R] Act as a senior product strategist with 10+ years of experience
in B2B SaaS.

[O] Help me outline a quarterly product roadmap presentation.

[K] The outline should have 6 main sections, each with 3-5 bullet points.
Keep it focused and executive-friendly — no more than 800 words total.

[E] The format should be something like:
I. Market Context
   - Key industry trends
   - Competitive landscape shifts
II. Quarterly Goals
   - Primary objective
   - Success metrics
...

See how different that is from "help me write a presentation"? The AI isn't guessing anymore. It knows who it's supposed to be, what you need, and what the output should look like.

The 12 Templates

1. Professional Article Writing

I use this one almost daily. The key is specifying the audience — that's what separates a generic article from one that actually resonates.

Act as a [field] expert and write a professional article about [topic].

Requirements:
1. Structure: introduction, 3-5 main sections, conclusion
2. Support each point with concrete examples or data
3. Write for [specific audience] — adjust depth and jargon accordingly
4. Approximately [X] words
5. Tone: [conversational / formal / technical]

What I've learned: "professional yet accessible" is not a useful instruction. Tell the AI exactly who's reading. "Write for marketing managers with 3 years of experience" produces much better results than "write for professionals."

2. Code Review and Debugging

This is the template that made me stop dreading code reviews.

Role: Senior [language] developer with [X] years of experience

I'm working on [brief project context] and I'm stuck on this:

[paste code]

Please:
1. Identify any bugs or potential issues
2. Provide a corrected version with comments explaining each change
3. Suggest 3 ways to improve performance or readability
4. Flag any security concerns

The trick is step 4. Most code review prompts miss security entirely. Adding it has caught issues I would have missed.

3. Data Analysis

I'm not a data scientist, so this template is basically my workaround.

Act as a data analyst. Here's my data: [paste or describe]

Please analyze this across these dimensions:
1. Overall trends — what's going up, what's going down?
2. Anomalies — anything unexpected or concerning?
3. Comparisons — how does this period compare to the last one?
4. Recommendations — what should I do about what I'm seeing?

Format: structured report with clear sections. Keep it actionable.

I used to ask "analyze this data" and get a wall of text. Specifying the four analysis dimensions changed the output completely.

4. Creative Copywriting

You're a senior copywriter specializing in [style].

Create [number] versions of copy for [product/service].
Channel: [social media / email / landing page / ad]

Each version should:
1. Be no more than [X] words
2. Highlight: [core selling points]
3. Target: [audience description]
4. Tone: [playful / urgent / sophisticated / warm]
5. End with a clear call to action

Why multiple versions? Because the first one is rarely the best. Generating 3-5 options and mixing and matching the best parts is much faster than iterating on a single draft.

5. Meeting Notes to Minutes

This one saves me 30 minutes every meeting.

Convert these rough meeting notes into formal meeting minutes:

[paste notes]

Structure:
1. Meeting info (date, attendees, agenda)
2. Key discussion points (3-5 per topic)
3. Decisions made (who decided what)
4. Action items (who does what, by when)
5. Next meeting details

Format: clean, scannable, with key decisions bolded.

The secret is "who does what, by when." Without that, meeting minutes are just a record of conversations, not a driver of action.

6. Learning Path Design

I want to learn [skill/topic]. My current level: [beginner / intermediate].
I can dedicate [time per week] and want to be proficient in [timeframe].

Create a structured learning plan:
1. Weekly milestones with specific learning objectives
2. Core topics to cover each week
3. Recommended resources (books, courses, projects)
4. A practical project for each phase to apply what I've learned
5. Common mistakes to avoid at my level

The practical project component is what makes this template work. Learning without application is just consumption. Every phase needs a project.

7. Problem Diagnosis

Here's my problem: [detailed description]
Context: [what's already been tried, current environment]

Please:
1. Identify the most likely root cause (use the 5 Whys approach)
2. Provide 3 solutions ranked by effort-to-impact ratio
3. For each solution: pros, cons, and rough implementation steps
4. Suggest how to prevent this from happening again

The "5 Whys" instruction is what elevates this from "here are some random suggestions" to actual problem-solving. The AI digs deeper instead of stopping at surface-level fixes.

8. Email Writing

Help me write a [type of email]
To: [relationship to recipient]
Subject: [topic]
Key message: [what I need to communicate]

Requirements:
1. Tone: [formal / casual / urgent / diplomatic]
2. Structure: clear opening, key points, specific ask
3. Include a clear deadline or next step
4. Keep it under [X] words
5. Close appropriately for the relationship

I used to spend 10-15 minutes on every email. This template gets me to a solid draft in 30 seconds. The key insight: specifying the relationship to the recipient ("my manager," "a potential client," "a colleague I don't know well") dramatically changes the tone.

9. Resume Optimization

Here's my current resume: [paste]

I'm applying for: [target role]

Please optimize it by:
1. Rewriting project experience using the STAR method
2. Quantifying achievements wherever possible
3. Highlighting skills that match this specific role
4. Reordering sections to lead with my strongest qualifications
5. Adjusting language to this industry's conventions

Output: optimized resume + a summary of what you changed and why.

The "summary of what you changed" is crucial. It teaches you what recruiters and hiring managers actually look for, so you get better at presenting yourself over time.

10. Brainstorming

Brainstorm around [topic]. I need at least [number] ideas.

For each idea, include:
1. Core concept (one sentence)
2. How it could work in practice
3. What makes it different from the obvious approach

Then rank them by:
- Innovation (1-10)
- Feasibility (1-10)
- Potential impact (1-10)

Recommend the top 3 with reasoning.

The ranking step is what separates brainstorming from idea vomiting. Without it, you get 20 undifferentiated ideas and no way to prioritize.

11. Translation and Localization

Translate this into [target language]: [text]

Requirements:
1. Preserve technical terminology accurately
2. Adapt expressions for [target language] cultural context
3. Adjust for [target audience] — don't just translate, localize
4. Keep the original formatting
5. Note any terms that don't translate cleanly and suggest alternatives

"Localize, don't just translate" is the key phrase here. It tells the AI to think about cultural fit, not just word-for-word accuracy.

12. Knowledge Extraction

Summarize and extract the key insights from this: [paste content]

Please provide:
1. Core takeaways (3-5 points)
2. Key concepts explained in plain language
3. A logical framework or mental model
4. How I can actually apply this
5. Questions this raises that are worth exploring further

Keep it concise. I should be able to review this in under 2 minutes.

The "under 2 minutes" constraint forces the AI to prioritize. Without it, you get a wall of text that's no faster to read than the original.

Why These Templates Work

Looking at all 12, there are a few common principles:

1. Specificity. Every template tells the AI exactly what you need. No ambiguity, no room for guessing.

2. Structure. Each prompt defines what the output should look like. The AI doesn't have to decide on format — you've already decided for it.

3. Context. Background information helps the AI calibrate. "I'm a PM at a startup" produces different output than "I'm a senior director at a Fortune 500."

4. Constraints. Word limits, format requirements, and "don't do X" instructions all reduce the AI's tendency to ramble.

5. Iterative design. These templates are meant to be starting points. The first output is rarely the final one — but it's usually 70% of the way there.

How to Use These

Don't try to memorize all 12. Pick the 2-3 that match your most frequent tasks, customize them for your specific needs, and save them somewhere accessible. Over time, you'll naturally develop your own variations.

The goal isn't to use these exact templates forever — it's to internalize the principles behind them so you can write effective prompts for any situation without needing a template at all.