AI-assisted design
Making design judgment visible
Experienced designers develop intuition for translating human insight into product decisions, but much of that reasoning remains difficult to explain, share, or repeat. I created the Value Activation Model (VAM) to make my own design reasoning explicit and use AI to translate behavioral research into actionable design recommendations.
AI-assisted design
Making design judgment visible
Experienced designers develop intuition for translating human insight into product decisions, but much of that reasoning remains difficult to explain, share, or repeat. I created the Value Activation Model (VAM) to make my own design reasoning explicit and use AI to translate behavioral research into actionable design recommendations.
Problem space
Design judgment is often a black box
Designers routinely translate research and experience into solutions, but the reasoning behind those decisions often remains implicit. This makes it difficult for teams to examine assumptions, communicate their rationale, or build on one another’s judgment.
Relevant knowledge is fragmented
Human values, emotions, behavioral science, and design patterns offer complementary perspectives on product experiences. Yet these bodies of knowledge are rarely integrated into a practical process that designers can apply to a specific challenge.
AI needs more than generative capability
AI can rapidly generate plausible design ideas, but those recommendations aren’t necessarily grounded in reliable research or a coherent understanding of human motivation. I wanted to combine AI’s ability to navigate knowledge with an explicit reasoning architecture.
Problem space
Design judgment is often a black box
Designers routinely translate research and experience into solutions, but the reasoning behind those decisions often remains implicit. This makes it difficult for teams to examine assumptions, communicate their rationale, or build on one another’s judgment.
Relevant knowledge is fragmented
Human values, emotions, behavioral science, and design patterns offer complementary perspectives on product experiences. Yet these bodies of knowledge are rarely integrated into a practical process that designers can apply to a specific challenge.
AI needs more than generative capability
AI can rapidly generate plausible design ideas, but those recommendations aren’t necessarily grounded in reliable research or a coherent understanding of human motivation. I wanted to combine AI’s ability to navigate knowledge with an explicit reasoning architecture.
Solution
An AI-assisted system for translating human values into design decisions
VAM connects four layers of design reasoning: Value → Emotion → Behavioral Principle → Design Pattern.
Each layer builds on the previous one, creating a traceable path from what matters to people to concrete recommendations for how a product should respond.
Three AI skills support different design activities: Scanner interprets qualitative research to surface underlying values, Auditor evaluates existing experiences to identify the values they express or overlook, and Advisor translates target values into emotional goals, behavioral principles, and recommended design moves.
Solution
An AI-assisted system for translating human values into design decisions
VAM connects four layers of design reasoning: Value → Emotion → Behavioral Principle → Design Pattern.
Each layer builds on the previous one, creating a traceable path from what matters to people to concrete recommendations for how a product should respond.
Three AI skills support different design activities: Scanner interprets qualitative research to surface underlying values, Auditor evaluates existing experiences to identify the values they express or overlook, and Advisor translates target values into emotional goals, behavioral principles, and recommended design moves.
Mind-blowing… it put an invisible thing into words that I can easily understand as a designer.
Senior Product Designer
Mind-blowing… it put an invisible thing into words that I can easily understand as a designer.
Senior Product Designer
So much of taste is instinctive and lives in silos. This lets teams like ours not just impulsively communicate why we reach decisions, but actually codify and socialize it together.
Head of Design
So much of taste is instinctive and lives in silos. This lets teams like ours not just impulsively communicate why we reach decisions, but actually codify and socialize it together.
Head of Design
My approach
My approach
Externalizing a decade of design reasoning
My experience across product design, human values, and behavioral science had given me a way of reasoning from people’s motivations to product decisions. I mapped that implicit process into four connected layers, combining established research with a structure designers could follow and inspect.
Externalizing a decade of design reasoning
My experience across product design, human values, and behavioral science had given me a way of reasoning from people’s motivations to product decisions. I mapped that implicit process into four connected layers, combining established research with a structure designers could follow and inspect.
Four-stage design reasoning model
Four-stage design reasoning model
Designing AI skills for reliable, actionable recommendations
I developed three AI skills with defined reasoning instructions, reference knowledge, and output structures. Using repeated test inputs, I compared their responses and refined the instructions to reduce unsupported conclusions, limit excessive recommendations, and preserve room for human judgment.
Designing AI skills for reliable, actionable recommendations
I developed three AI skills with defined reasoning instructions, reference knowledge, and output structures. Using repeated test inputs, I compared their responses and refined the instructions to reduce unsupported conclusions, limit excessive recommendations, and preserve room for human judgment.
AI skills with structured reasoning and reference libraries
AI skills with structured reasoning and reference libraries
Applying the model to real product experiences
I tested VAM against existing digital products, examining how their design decisions supported or frustrated particular human values. These audits helped make the reasoning tangible, showing how the model could move beyond general UX advice toward more specific, values-informed recommendations.
Applying the model to real product experiences
I tested VAM against existing digital products, examining how their design decisions supported or frustrated particular human values. These audits helped make the reasoning tangible, showing how the model could move beyond general UX advice toward more specific, values-informed recommendations.
AI-assisted product audit of Spotify
AI-assisted product audit of Spotify
Translating the system into a practitioner experience
Using AI-assisted development, I created the public website, toolkit, visual identity, and workshop materials. I’ve delivered multiple public practitioner workshops, translating the theoretical model into accessible explanations, examples, and exercises that designers could use in their own work.
Translating the system into a practitioner experience
Using AI-assisted development, I created the public website, toolkit, visual identity, and workshop materials. I’ve delivered multiple public practitioner workshops, translating the theoretical model into accessible explanations, examples, and exercises that designers could use in their own work.
VAM practitioner workshop and exercises
VAM practitioner workshop and exercises
Outcomes and learning
An original, publicly available AI methodology
I brought VAM from an internal reasoning process into a documented framework, three reusable AI skills, a public website, and a practitioner workshop. The system makes an approach to design judgment accessible beyond my own practice.
Real-world practitioner exposure
Early practitioner feedback highlighted the value of making intuitive design reasoning explicit and giving teams a shared language for discussing the decisions behind their work.
A foundation for further validation
The project gave me practical experience developing AI-assisted tools and using AI throughout the product development process, from defining system behavior to building and iterating on working experiences.
VAM remains an evolving methodology. Its ability to improve downstream product outcomes has not yet been formally validated.
Outcomes and learning
An original, publicly available AI methodology
I brought VAM from an internal reasoning process into a documented framework, three reusable AI skills, a public website, and a practitioner workshop. The system makes an approach to design judgment accessible beyond my own practice.
Real-world practitioner exposure
Early practitioner feedback highlighted the value of making intuitive design reasoning explicit and giving teams a shared language for discussing the decisions behind their work.
A foundation for further validation
The project gave me practical experience developing AI-assisted tools and using AI throughout the product development process, from defining system behavior to building and iterating on working experiences.
VAM remains an evolving methodology. Its ability to improve downstream product outcomes has not yet been formally validated.
Challenges and lessons
Making abstract reasoning actionable
The challenge was preserving the intellectual foundations of the model without overwhelming practitioners. I worked to make the reasoning understandable through a simple sequence, concrete product examples, and recommendations that designers could immediately evaluate.
Supporting judgment rather than replacing it
I designed VAM to inform decisions rather than make them autonomously. Limiting recommendations, exposing their underlying rationale, and accommodating different design activities helped keep the system useful without positioning AI as a substitute for human expertise.
Challenges and lessons
Making abstract reasoning actionable
The challenge was preserving the intellectual foundations of the model without overwhelming practitioners. I worked to make the reasoning understandable through a simple sequence, concrete product examples, and recommendations that designers could immediately evaluate.
Supporting judgment rather than replacing it
I designed VAM to inform decisions rather than make them autonomously. Limiting recommendations, exposing their underlying rationale, and accommodating different design activities helped keep the system useful without positioning AI as a substitute for human expertise.