Design Thinking
What problem are you actually solving?
Design thinking is problem-solving through the lens of human needs. Not what the system requires — what the person requires. The Dreamer archetype uses it to turn possibility into vision.
The Stages
Five stages, but never linear. You loop back constantly — the test reveals what empathy missed.
| Stage | Question | Technique | Output |
|---|---|---|---|
| Empathize | What do they actually need? | Journey mapping, observation | Pain points, mental models |
| Define | What's the real problem? | Reframing, constraint mapping | "How might we..." statement |
| Ideate | What could solve this? | Crazy eights, reverse brainstorm | 30+ raw concepts |
| Prototype | What's the smallest test? | Paper sketches, service blueprint | Tangible representation |
| Test | Does it actually work? | Five-second test, A/B testing | Evidence, not opinion |
The Feedback Loop
Design thinking is a control system. Each stage creates feedback that corrects the next:
EMPATHIZE → DEFINE → IDEATE → PROTOTYPE → TEST
↑ |
└──────── what you learned ──────────────┘
| Control Concept | Design Equivalent | What Goes Wrong Without It |
|---|---|---|
| Setpoint | User's desired outcome | You optimize the wrong thing |
| Sensor | User testing, telemetry | You design blind |
| Actuator | Prototype iteration | You ship assumptions |
| Lag | Time between action and feedback | Oscillation — over-correcting |
Shorter feedback loops produce better designs. Paper prototype in 30 minutes beats pixel-perfect mockup in 2 weeks — because you learn faster.
Design Affordance
Agency is the capacity to act. Affordance is the design that makes acting on it obvious. Together: design that affords agency makes the right action visible, doable, and worth doing.
The term comes from psychologist James Gibson and was carried into product design by Don Norman in The Design of Everyday Things. Affordances are the perceived possibilities for action a system offers — a door handle says pull, a flat plate says push, a slider says drag. The form suggests the function before any instruction.
| Layer | Bad design | Affordance |
|---|---|---|
| Visibility | Hidden control | Obvious cue |
| Mapping | Unclear cause and effect | Direct match |
| Feedback | Silent action | Confirmed result |
| Constraints | Dangerous freedom | Guided rails |
| Conceptual model | Confusing mental model | Matches user expectation |
AI Affordance Layer
AI does not just add capabilities. When designed well, it affords them — making the right action visible, contextual, and doable inside the user's existing flow. When designed badly, it adds capability without affordance — capability without affordance is friction wearing a clever hat.
The test: after someone uses your AI, do they know one more thing they can do — or do they have one more thing to figure out? If the answer is the second one, the design has added agency without affordance.
Demand → Design → Affordance
Affordance only matters when it serves a real action. Validating product demand is upstream: see Validate Demand for the framework that turns pain into a named, paid-for, deliverable job. Design thinking then turns that validated job into affordance — the visible path the user actually takes.
| Stage | What it produces | Without it |
|---|---|---|
| JTBD validate-demand | A real action worth designing for | Solutions to imagined pain |
| Design thinking | Affordance — the path made obvious | Capability the user never finds |
| Build + ship | The product runs the loop | Affordance that never reaches anyone |
Cut any of the three and the loop breaks. The full thinking-line: see AI Amplifies Agency for how affordance, agency, and demand compound.
Named Frameworks
| Framework | Originator | Model | Best For |
|---|---|---|---|
| Double Diamond | British Design Council | Diverge → Converge → Diverge → Converge | Strategic projects |
| Design Sprint | Google Ventures | 5-day intensive: Map → Sketch → Decide → Prototype → Test | Rapid validation |
| Jobs to Be Done | Christensen | What "job" is the user hiring this product to do? | Product positioning |
| Lean UX | Gothelf | Hypothesis → MVP → Measure → Learn | Agile product teams |
Techniques
Empathize
| Technique | How | Time |
|---|---|---|
| Journey Mapping | Timeline of user experience with emotional peaks and valleys | 1-2 weeks |
| Empathy Mapping | What user says, thinks, feels, does | 1-2 hours |
| Contextual Observation | Watch users in their environment — what they do vs what they say | 1-2 weeks |
Ideate
| Technique | How | Time |
|---|---|---|
| Crazy Eights | Fold paper into 8 sections, sketch 8 ideas in 8 minutes | 15 minutes |
| Reverse Brainstorm | "How could we make this worse?" Invert to find principles | 1 hour |
| Analogous Transfer | How does nature/another industry solve this? | 1-2 hours |
Test
| Technique | What It Reveals | Sample |
|---|---|---|
| Five-Second Test | First impression, headline clarity | 20-50 people |
| Moderated Testing | Where users struggle, what delights | 5-8 people |
| A/B Testing | Which variant performs better | 100+ users |
The Integration
Design thinking alone optimizes for user delight. Systems thinking alone optimizes for the whole. Together:
| Alone | Integrated |
|---|---|
| User loves the feature | Feature works within the system |
| Elegant solution in isolation | Solution doesn't create perverse incentives |
| Sprint-based, short horizon | Tests against future scenarios |
| Empathy for end-users only | Includes stakeholders, operators, policy |
Matrix thinking makes the gaps visible. Design thinking fills them with solutions. First principles ensures the solutions stand on truth.
Context
- Control System — Design thinking is a control system with human needs as setpoint
- Product Design — The measurable standards for design output
- Design Review — The audit checklist for design work
- Dreamer Archetype — The mindset that runs this method
- Matrix Thinking — Makes gaps visible for design to fill
Questions
What are you designing for — the user's stated need or the need they can't articulate?
- At which stage of empathize → define → prototype are you most likely to shortcut — and why?
- When a prototype fails, is it evidence against the solution or evidence against the problem definition?
- What human need is your current product solving that you didn't know about when you built it?