← BACK

LAB / EXPERIMENT 02

AI Workflows

I've always thought by making — sketching, diagramming, prototyping and giving an idea enough form to react to it.

AI has dramatically shortened the distance between those first marks and something I can actually examine.

EXPLORATION · PROTOTYPING · AI · SPATIAL DESIGN · HCI

THE QUESTION

What happens when making becomes almost instantaneous?

The tools are getting faster. An idea that once took days or weeks to visualize can now take hours — sometimes minutes.

I wanted to understand what that changes about the designer's role.

Watercolor site plan sketch of a small dark cabin in a clearing surrounded by forest, with water to the north, trails to the southwest and a vehicle access road leading up from the south.
01 · SKETCHAPPLE PENCIL · IPAD

THE HYPOTHESIS

Faster tools make human judgment more important, not less.

Before there is a prompt, there is still a question.

I sketch to understand the problem, give it constraints, move things around and discover what I don't know. AI enters after that — helping me explore possibilities faster and giving me something more tangible to question.

SKETCH
02 · STRUCTUREAUTOCAD REVUP · DIMENSIONED PLAN
  1. QUESTION
  2. SKETCH
  3. STRUCTURE

01 / RECOVERY ROOST

From a rough sketch to a place I could walk through in my head.

Recovery Roost began as a spatial question: how might a very small home be designed around recovery, wellness and connection to nature?

I worked through the idea by hand first — sketching layouts, separating public and private space, thinking through movement, constraints and the relationship between the home and its environment.

Then I used digital and AI-assisted tools to push the idea further: testing spatial arrangements, visualizing materials and atmosphere, and seeing how the concept felt once it became a place instead of a diagram.

Design concept diagram in four steps: a 25 square meter maximum base volume, a spatial grid of modules, the volume split into private above and public below, and rooms and levels numbered for bedroom, sauna, bathroom, cold plunge and patio, kitchen, living and dinette.
03 · DEFINE + CONSTRAINAUTOCAD REVUP · ADOBE ILLUSTRATORFOOTPRINT · SPATIAL LOGIC · PRIVACY · PROGRAM
AI-generated interior of the compact living space: a pale linen sectional and ottoman on a light wood floor, with a small kitchen, open shelving and a glass door looking out to the deck.
AI-generated cutaway of the two-level structure, with a lofted bedroom under a timber ceiling above a living room and kitchen on the ground floor.
AI-generated recovery environment: a timber sauna cabin and sunken cold plunge on a wood deck beside a lake and pine forest.
AI-generated exterior of the board-and-batten structure with large windows and floating stairs, set among tall pines beside a lake.
04 · VISUALIZE + EVALUATEGPT · AI-ASSISTED VISUALIZATIONSPACE · MATERIAL · ATMOSPHERE · RECOVERY · NATURE
  1. SKETCH
  2. STRUCTURE
  3. CONSTRAIN
  4. GENERATE
  5. EVALUATE
  6. REFINE

02 / PETITE PALACE

Then I used the same tools to imagine something that didn't exist at all.

Petite Palace started with a different question: what if technology could change our perception of how much space we have?

Instead of designing only within physical square footage, I explored virtual home extensions — spaces that could expand the experience of home through immersive technology.

AI and emerging visualization tools helped me move quickly between product thinking, spatial design, interface concepts and speculative environments.

A woman wearing a VR headset and headphones in a room lit pink and blue, smiling with both hands raised as she explores a virtual space.
01 · IMAGINE
02 · REPRESENT
03 · EXPERIENCE
04 · QUESTION
  1. IMAGINE
  2. REPRESENT
  3. EXPERIENCE
  4. QUESTION

The value wasn't simply producing more images.

It was being able to make an unfamiliar idea visible enough to interrogate.

WHAT CHANGED

My workflow is becoming less linear.

I used to think of the process as moving from idea to design to prototype.

Now it behaves more like a loop.

  1. WONDER
  2. SKETCH
  3. MAKE
  4. REACT
  5. QUESTION
  6. CHANGE
  7. MAKE AGAIN

Each pass makes the idea more specific. AI lets me move through those passes faster — but speed only matters if I'm still willing to stop, look and decide what should change.

WHAT I LEARNED

The bottleneck is moving from making to deciding.

AI can help generate an image, explore a spatial concept, prototype an interface or turn an idea into working software remarkably quickly.

But it still needs someone to ask the question, frame the problem, recognize what isn't working and decide what deserves to become real.

The tools got faster. The thinking still starts here.

make → look → question → make again

CURRENTLY EXPLORING

CLAUDE · GEMINI · MCP · RAPID PROTOTYPING · IMAGE GENERATION · HUMAN–AI WORKFLOWS

NEXT EXPERIMENT

Health, in motion.

Exploring how personalized data, behavioral feedback, and spatial computing could turn healthy choices into learned habits.