Design Thinking with Prompt Patterns

Using the rule of three, I've come up with groups of prompt patterns that can be used in the design thinking process for problem solving.

This is a high level approach that serves as a framework with which to get started. As you get deeper into the design thinking toolkit, additional prompts can be added to your repertoire. Here are some prompts matched to each of the five stages of the design thinking process.

Empathise

Goal: Understand users through research, interviews, and observation.

Flipped Interaction Pattern

The Flipped Interaction Pattern reverses the typical AI-human dynamic. Instead of the users asking the LLM questions, this pattern empowers users to step into different roles while the LLM asks targeted, empathy-building questions.

This pattern helps in the Empathise stage by:

• Simulating user interviews: The LLM poses as an interviewer, asking questions that uncover user needs, pain points, and motivations that might otherwise remain hidden.

• Role-playing exercises: Users can assume the perspective of their target audience while the LLM guides them through scenarios, helping reveal emotional responses and contextual challenges.

• Guided reflection: Rather than users asking the LLM questions, the LLM prompts users to examine assumptions and biases they might hold about their users.

Example prompt format:

I want you to interview me as if I'm [specific user type]. Ask me probing questions about my experiences with [relevant problem/product]. After 5-7 questions, summarize key insights and potential unmet needs you've identified.

This flipped dynamic creates space for deeper empathy development and often surfaces unexpected insights that traditional research methods might miss.

Persona Pattern

The Persona Pattern empowers you to build robust user personas with unprecedented depth and insight, serving as the foundation for empathetic design.

This pattern helps in the Empathise stage by:

• Creating realistic user profiles: The LLM can generate comprehensive personas based on demographics, behaviours, societal norms, and psychographics that represent your target users.

• Surfacing hidden needs: By having the LLM role-play as different personas, you can uncover motivations and pain points that might not emerge through traditional research methods.

• Challenging assumptions: The LLM can help validate or question your existing beliefs about users by providing evidence-based perspectives from diverse user viewpoints.

Example prompt format:

Create a detailed persona for a [specific user type] who would use [product/service]. Include demographic information, goals, frustrations, a typical day in their life, and their relationship with technology. Then, speaking as this persona, describe your experience with [specific task/problem] and what would make an ideal solution for you.

This pattern helps teams develop a shared understanding of users that goes beyond surface-level characteristics, creating a foundation for truly user-centered design decisions throughout the process.

Semantic Filter Pattern

The Semantic Filter Pattern helps extract meaning from complex information by filtering and organising it in meaningful ways, enhancing understanding of user feedback and research data.

This pattern helps in the Empathise stage by:

• Organising qualitative data: The LLM can process interview transcripts, survey responses, or field notes to identify recurring themes, emotional patterns, and key insights that might be missed by manual analysis.

• Reducing bias: By having the LLM filter information based on specific criteria, you can minimise confirmation bias and ensure a more objective understanding of user needs.

• Finding hidden connections: The LLM can identify non-obvious relationships between different user statements or behaviours, revealing deeper insights about their motivations and needs (functional and emotional).

Example prompt format:

Analyze the following [user research data/interview transcripts/survey responses]. Filter and organize the information to identify: 1) primary pain points, 2) unmet needs, 3) emotional responses, and 4) behavioural patterns. For each category, provide direct quotes or observations, and suggest what these might reveal about underlying user motivations.

This pattern transforms raw research data into structured insights, making it easier to develop a nuanced understanding of users and their contexts before moving into the Define stage.

DEFINE

Goal: Synthesize findings into clear, actionable problem statements.

Cognitive Verifier Pattern

The Cognitive Verifier Pattern helps validate and verify thinking processes, ensuring that problem definitions are accurate, comprehensive, and truly representative of user needs.

This pattern helps in the Define stage by:

• Scrutinizing problem statements: The LLM can analyse draft problem statements to identify logical flaws, unfounded assumptions, or biases that might undermine your understanding of user needs.

• Testing evidence connections: By verifying that your problem definitions are directly supported by research findings, the LLM ensures that solutions will address genuine user needs rather than perceived ones.

• Challenging framings: The LLM can suggest alternative ways to frame the problem that might open up different solution spaces or reveal overlooked aspects of user experiences.

Example prompt format:

Review this problem statement: "[problem statement]". Verify whether it: 1) is supported by the user research data, 2) contains hidden assumptions, 3) addresses the root cause rather than symptoms, and 4) is specific enough to guide ideation but open enough to allow creativity. Suggest refinements to improve the statement.

This pattern helps teams avoid the common pitfall of rushing to solve ill-defined problems, ensuring that the foundation for the rest of the design process is solid and truly user-centered.

Chain of Thought Prompting

The Chain of Thought Prompting pattern guides LLMs through explicit reasoning steps, helping refine problem definitions with greater clarity and precision.

This pattern helps in the Define stage by:

• Breaking down complex problems: By working through problems step-by-step, this pattern helps decompose complex user needs into manageable components that can be addressed systematically.

• Revealing hidden assumptions: The explicit reasoning process exposes assumptions that might otherwise remain implicit, allowing them to be examined and validated against research findings.

• Connecting insights to problem statements: This pattern creates clear logical pathways from research insights to problem definitions, ensuring that problem statements are firmly grounded in user needs.

Example prompt format:

Let's define our problem statement step-by-step:

1. First, summarize the key insights from our user research about [specific user group]

2. Next, identify the most significant pain points or unmet needs

3. Then, consider the contextual factors that influence these needs

4. Now, draft a "How Might We" statement that addresses the core problem

5. Finally, refine this statement to ensure it's specific, actionable, and user-centered

Walk through each step in detail, showing your reasoning at each stage.

This pattern improves the quality of problem definitions by making the reasoning process explicit, reducing the risk of logical leaps or unfounded conclusions that could lead to solving the wrong problem.

Fact Check List Pattern

The Fact Check List Pattern enables systematic verification of problem statements against evidence, ensuring accuracy and relevance before proceeding with solution development.

This pattern helps in the Define stage by:

• Validating problem framing: The LLM generates a comprehensive checklist to verify that problem statements accurately reflect user research findings and aren't based on unfounded assumptions.

• Ensuring comprehensiveness: This pattern helps identify gaps in problem definitions by systematically checking whether all key user needs and contextual factors have been considered.

• Prioritizing effectively: By evaluating problem statements against evidence-based criteria, teams can determine which problems are most important to solve first, based on user impact, business goals, and resource constraints.

Example prompt format:

Create a fact-checking framework to verify our problem statement: "[problem statement]". For each of the following criteria, evaluate whether our statement meets the standard and provide evidence from our research:

1. User-centricity: Does it focus on user needs rather than business/technical needs?

2. Evidence-based: Is it supported by specific research findings?

3. Specificity: Is it specific enough to guide ideation without prescribing solutions?

4. Impact: Does solving this problem meaningfully improve the user experience?

5. Actionability: Is it within our capability to address?

For any criteria where the statement falls short, suggest specific revisions.

This pattern creates a systematic verification process that helps teams avoid pursuing solutions to poorly defined problems, increasing the likelihood that subsequent design efforts will address genuine user needs.

IDEATE

Goal: Generate multiple, diverse, creative ideas.

Few-Shot Examples

The Few-Shot Examples pattern leverages examples to guide the LLM in generating diverse, creative solutions during ideation sessions.

This pattern helps in the Ideate stage by:

• Jumpstarting creativity: Providing a few examples helps overcome initial creative blocks by demonstrating the expected format and quality of ideas without constraining thinking.

• Ensuring diversity: By carefully selecting diverse examples, you can encourage the LLM to explore different solution spaces and approaches to the problem.

• Maintaining relevance: Examples help ground ideation in solutions that are relevant to the defined problem while still allowing for creative exploration.

Example prompt format:

Based on our problem statement and design criteria: [problem statement], generate 10 innovative solutions.

Here are three diverse examples of the kind of ideas I'm looking for:

1. [Example idea 1 that approaches the problem from perspective A]

2. [Example idea 2 that approaches the problem from perspective B]

3. [Example idea 3 that approaches the problem from perspective C]

Now generate 15 more ideas that are as diverse and creative as these examples, but explore different approaches. For each idea, explain how it addresses the core user need and what makes it innovative.

This pattern strikes a balance between providing guidance and enabling creative freedom, resulting in ideation sessions that produce both practical and innovative solution concepts.

Game Play Pattern

The Game Play Pattern transforms the ideation process into an engaging, structured game that stimulates creativity and breaks conventional thinking patterns.

This pattern helps in the Ideate stage by:

• Overcoming creative blocks: By framing ideation as a game with specific rules and constraints, participants can bypass mental barriers and generate ideas more freely.

• Encouraging divergent thinking: Game mechanics push participants to explore unconventional solution spaces they might otherwise avoid in traditional brainstorming.

• Building on ideas collaboratively: Games naturally create an environment where participants build upon each other's contributions, leading to more robust and innovative solutions.

Example prompt format:

Let's play an ideation game called "Impossible to Possible" for our problem: [problem statement]

Game rules:

Round 1: Generate 5 "impossible" solutions that seem completely unfeasible or even magical

Round 2: For each impossible idea, identify the core principle that makes it appealing

Round 3: Transform each impossible idea into a possible solution by finding realistic ways to implement its core principle

Final round: Combine elements from different solutions to create 3 innovative hybrid concepts

For each round, explain your thinking process and how the ideas relate to addressing user needs.

This pattern taps into the power of playfulness and structured constraints to produce breakthrough ideas that might not emerge through conventional ideation techniques.

Menu Action Pattern

The Menu Action Pattern provides a structured way to generate ideas by offering a menu of creative "actions" that can be applied systematically to a problem.

This pattern helps in the Ideate stage by:

• Systematizing creativity: By providing a clear menu of creative techniques or approaches, it helps teams move beyond conventional thinking and explore different angles systematically.

• Expanding solution spaces: Each "action" in the menu pushes ideation in a specific direction, ensuring teams explore a wider range of potential solutions than they might otherwise overlook.

• Overcoming fixation: When teams get stuck on particular solution types, the menu offers ready-made prompts to shift thinking and explore fresh approaches.

Example prompt format:

Let's generate ideas for solving our problem: [problem statement]

Apply each of the following idea generation techniques and generate 2-3 ideas for each:

1. Inversion: What if we solved the opposite problem? How could we then adapt that solution?

2. Cross-industry inspiration: How have similar problems been solved in [unrelated industry 1], [unrelated industry 2], and [unrelated industry 3]?

3. Constraint addition: What if we had to solve this with [specific constraint]?

4. Absurd exaggeration: What if we exaggerated the solution to an impossible degree, then scaled it back to reality?

5. Biomimicry: How might nature solve this problem?

6. Counter-logic: In what ways will this not work?

For each technique, explain how the resulting ideas address the core user need identified in our problem statement.

This pattern transforms ideation from an unstructured brainstorming session into a methodical exploration of different creative territories, resulting in a more diverse and comprehensive set of potential solutions.

PROTOTYPE

Goal: Build low-fidelity versions of solutions.

Recipe Pattern

The Recipe Pattern provides a clear, step-by-step framework for turning abstract ideas into tangible prototypes, similar to how cooking recipes break down complex dishes into manageable instructions.

This pattern helps in the Prototype stage by:

• Breaking down complexity: Transforms sophisticated design concepts into sequential steps, making prototype creation accessible even to team members without specialized design skills.

• Ensuring consistency: Creates a standardized approach to prototyping, allowing multiple team members to build compatible components that work together seamlessly.

• Facilitating iteration: Makes it easy to identify and modify specific elements of a prototype when feedback indicates the need for changes.

Example prompt format:

Create a prototype recipe for our solution concept: [solution concept]

Prototype Recipe

1. Materials needed: List all resources required (digital tools, physical materials, content elements)

2. Preparation steps: Outline what needs to be gathered or created before assembly begins

3. Core components: Break down the prototype into 3-5 essential components

4. Assembly instructions: Provide step-by-step directions for building each component

5. Integration guide: Explain how to combine components into a coherent prototype

6. Testing preparation: Include instructions for setting up the prototype for user testing

For each step, include estimated time requirements and potential challenges to anticipate.

This pattern transforms abstract ideas into concrete implementation plans, making prototyping more efficient and accessible while ensuring that the resulting prototypes effectively embody the core elements of the solution concept.

Template Pattern

The Template Pattern provides pre-designed structural frameworks that help transform abstract ideas into consistent, user-testable prototypes with minimal effort.

This pattern helps in the Prototype stage by:

• Standardizing representation: Offers pre-designed frameworks that ensure consistent visualization of ideas, making prototypes more professional and easier to evaluate.

• Accelerating creation: Reduces the time and effort needed to build functional prototypes by providing ready-made structures that only need to be populated with solution-specific content.

• Focusing on key elements: Templates direct attention to the critical components that need to be prototyped, preventing teams from getting lost in details that aren't essential for testing.

Example prompt format:

Generate a prototype template for our solution concept: [solution concept]

Create a structured template with the following elements:

1. User interface framework: Provide a skeletal structure for the main screens/touchpoints

2. User flow diagram: Map out the key interaction pathways

3. Content placeholders: Include placeholders for critical information and functionality

4. Variable elements: Identify which elements should be customizable for different testing scenarios

5. Annotation guide: Instructions for how to document user interactions and feedback during testing

Design this template to be implementation-agnostic (usable for paper, digital, or hybrid prototypes) while still capturing the essence of our solution concept.

This pattern bridges the gap between ideation and testing by providing structured frameworks that make prototype creation more efficient while ensuring that the resulting prototypes effectively communicate the core value proposition to users.

Meta Language Creation Pattern

The Meta Language Creation Pattern creates specialized vocabularies or frameworks to articulate and prototype complex ideas, making abstract concepts more tangible and testable.

This pattern helps in the Prototype stage by:

• Bridging abstraction gaps: Develops specialized terminology that helps translate abstract concepts into concrete prototypes, ensuring team alignment on what's being built.

• Creating consistent representations: Establishes a shared language for describing prototype elements, improving communication and reducing misinterpretations during development.

• Enabling systematic exploration: Provides a structured way to map out the various dimensions of a solution space, helping teams create more comprehensive prototype variations.

Example prompt format:

For our solution concept: "[solution concept]", let's create a meta-language to prototype effectively.

1. Core vocabulary: Define 5-7 key terms that capture the essential elements of our solution

2. Relationship syntax: Create rules for how these elements interact with each other

3. User interaction grammar: Develop ways to express how users will engage with the solution

4. State changes: Define terminology for how the system transforms through user interactions

5. Visual/physical representation: Create a notation system for representing these elements in prototypes

Apply this meta-language to create a prototype specification that team members can use to build consistent representations of our solution for testing.

This pattern is particularly valuable for prototyping complex systems or solutions with multiple interconnected components, creating a shared foundation that ensures all prototype elements work together coherently.

TEST

Goal: Collect feedback and evaluate prototypes.

LLMs Grade Each Other

The LLMs Grade Each Other pattern leverages multiple AI models to evaluate prototypes and generate comprehensive, multi-perspective feedback without human bias.

This pattern helps in the Test stage by:

• Eliminating confirmation bias: By having multiple LLM "reviewers" assess prototypes independently, teams can get feedback that isn't influenced by the creator's attachment to their solution.

• Simulating diverse user perspectives: Different LLM instances can be prompted to evaluate from various stakeholder viewpoints, providing a more comprehensive understanding of how different users might react.

• Identifying blind spots: When multiple models evaluate the same prototype using different criteria, they often surface issues or opportunities that might be missed by a single evaluator.

Example prompt format:

Let's evaluate our prototype for: [prototype description]

I'll have three different LLM evaluators assess this prototype:

Evaluator 1: Take on the perspective of our primary user persona [persona details]. Evaluate this prototype based on how well it meets their specific needs and expectations. Focus on usability and value proposition.

Evaluator 2: Analyze this prototype from a technical feasibility standpoint. Identify potential implementation challenges, scalability issues, and technical limitations that might affect the user experience.

Evaluator 3: Evaluate this prototype through the lens of business viability. Consider cost factors, market differentiation, and potential barriers to adoption.

Each evaluator should:

1. Provide a rating from 1-10 for the prototype on their assigned dimension

2. Identify 3 specific strengths from their perspective

3. Highlight 3 areas for improvement

4. Suggest 2-3 concrete modifications that would address the identified issues

Finally, synthesize the feedback from all three perspectives into a prioritized list of revisions for the next prototype iteration.

This pattern transforms testing from a potentially subjective exercise into a structured, multi-dimensional evaluation that captures diverse perspectives, helping teams identify the most critical improvements to make before the next design iteration.

Menu Action Pattern

The Menu Action Pattern provides a systematic approach to testing prototypes by offering a structured menu of evaluation techniques or perspectives.

This pattern helps in the Test stage by:

• Providing comprehensive feedback: By applying multiple evaluation methods systematically, teams can gather more thorough feedback than with a single testing approach.

• Ensuring balanced assessment: The menu ensures teams don't over-focus on certain aspects of testing (like usability) while neglecting others (like technical feasibility).

• Standardizing evaluation: Creates consistency in how different prototypes are assessed, making it easier to compare solutions objectively.

Example prompt format:

Let's evaluate our prototype for: [prototype description]

Apply each of these testing approaches and document the findings:

1. Usability assessment: Evaluate how intuitive and accessible the prototype is for first-time users

2. Value proposition test: Assess how effectively the prototype communicates and delivers on its core value

3. Edge case exploration: Identify how the prototype handles unexpected user behaviors or inputs

4. Competitive comparison: Compare the prototype against existing solutions on 3-5 key dimensions

5. Emotional response mapping: Document the emotional journey users might experience while using the prototype

For each testing approach:

- Document key observations

- Rate performance on a scale of 1-5

- Identify specific improvement opportunities

- Prioritize changes for the next iteration

This structured evaluation will ensure we gather comprehensive feedback that addresses all critical aspects of our solution.

This pattern transforms testing from a potentially subjective or one-dimensional process into a methodical evaluation across multiple critical dimensions, resulting in more actionable insights for prototype refinement.

Tail Generation Pattern

The Tail Generation Pattern extends scenarios beyond their initial endpoints, revealing hidden assumptions, edge cases, and consequences that might otherwise be overlooked during testing.

This pattern helps in the Test stage by:

• Uncovering hidden assumptions: By extending test scenarios beyond their expected conclusions, teams can identify implicit assumptions about how users will interact with the solution.

• Exploring edge cases: Prompts the team to consider what happens in unusual or extreme usage scenarios, revealing potential weaknesses or opportunities for robustness.

• Identifying unintended consequences: Helps teams anticipate potential ripple effects or long-term implications of their solution that might not be apparent in standard testing.

Example prompt format:

For our prototype: "[prototype description]", let's explore what happens beyond the primary user journey.

Generate "tail scenarios" by extending these test cases beyond their expected endpoints:

1. Standard usage scenario: [describe typical use case]

→ What happens after the user completes this interaction?

→ What are 3 potential follow-up actions users might take?

→ What expectations might users develop for future interactions?

2. Edge case scenario: [describe unusual use case]

→ What are the downstream effects if this edge case occurs?

→ How might this affect user trust or system integrity?

→ What secondary problems could emerge from this situation?

3. Error recovery scenario: [describe situation where user encounters a problem]

→ After the user resolves this error, how might their behavior change?

→ What lingering perceptions might affect their future usage?

→ What compensating behaviors might they develop?

For each tail scenario, identify:

- Hidden assumptions in our prototype design

- Potential improvements to make the solution more robust

- Opportunities to create better continuity in the user experience

This pattern transforms testing from a discrete set of isolated scenarios into a continuous exploration of the user journey, helping teams develop solutions that hold up better in real-world conditions where usage rarely follows idealized paths.

CROSS-STAGE SUPPORT

Question Refinement Pattern

Use anytime to improve vague user prompts, reframe problem statements, or make ideation inputs clearer.

Combining Patterns

Design your own prompt stacks dynamically as your process evolves — combining patterns is encouraged!