Enter as a beginner
I used the integrated workspace as a first-time startup founder and recorded every hesitation instead of explaining it away.
01 · OVERVIEW ONE MONTH · TWO PEOPLE · THREE LIVE PRODUCTS
Vetra AI · Product strategy, research, design, and build
One product was hiding three jobs.
I joined a technical founder as the only designer. In one month, I turned first-time confusion into a product strategy—and carried that strategy through research, three focused landing pages, a redesigned workspace, and production launch.
RoleProduct designer + design builder
TeamTechnical founder + me
Duration1 month
Outcome3 focused products, live
02
WEEK 1 · UNDERSTAND
I began by preserving my own confusion.
··········›
WHAT DOES THIS DO?
WHAT HAPPENS NEXT?
Capture friction
Each question became a concrete note attached to the moment where value was interrupted.
··········›
Make it shared
The founder answered each friction point, exposing the gap between the builder’s internal model and a new user’s view.
“The builder saw one coherent system. First-time users saw unanswered questions before value.”
03
WEEK 1 · IMPROVE
We fixed the obvious friction before asking the market a bigger question.
The scan was not a usability report left on a shelf.
AI helped cluster repeated questions and propose possible responses. The founder supplied system context. I decided which recommendations were useful, what should change first, and where we still needed evidence.
- Observed behavior remained the source.
- AI accelerated organization—not product judgment.
- The improved product became the research stimulus for the field study.
OBSERVEDFirst-time friction
EXPLAINEDFounder intent
SYNTHESIZEDRepeated patterns
DECIDEDWhat to change now
04
WEEK 2 · LEARN
With limited analytics, I went where the users were.
Why field research?The product was too early for reliable behavioral volume. Direct observation made invisible hesitation visible, while the structured questionnaire made responses comparable.
05
WEEK 2 · SYNTHESIZE
I used AI to interrogate the dataset. Then start human judgment.
THE FINDING
Users were blocked before discovering features that could have been valuable to them.
06
WEEK 2 · DIAGNOSE
Users stopped before discovering potential value.
ENTEROne unfamiliar system
FRICTION“Where am I?”
HIDDENSlides
HIDDENVideo
HIDDENEdit Everything
Working assumption: because this product model was unfamiliar, understanding it required too much effort. Confusion reduced motivation before users reached the core capabilities.
07
WEEK 2 · REFRAME
One product was trying to serve three core roles across different purposes and moments.
Explain the product at a high level
Turn interface screens into a demo video for people who do not understand the product yet.
Demo videoMake a complex idea understandable
Break dense information into bite-sized visual explanations and presentation-ready slides.
Visual slidesRegain control of a flat visual
Upload an image, separate it into editable layers, then move, delete, or rewrite any element.
Edit Everything08
WEEK 2 · DECIDE
The strategic fork.
Keep one integrated product
Find a new mental model that makes three capabilities feel coherent.
- Preserves the combined technical moat
- Could create deeper attachment for multi-job users
- Requires redesign, engineering, recruitment, and repeated testing
- Feedback remains difficult to attribute
OR
Split into three focused products
Give each audience one clear entrance, promise, and job.
- Removes discovery complexity immediately
- Creates cleaner role-specific learning
- Reduces early-stage development risk
- Requires three positioning and acquisition strategies
09
THE ARGUMENT
I chose the option that improved the quality and speed of learning.
The expensive ambiguity loop
- Invent a new mental model
- Re-engineer the integrated workspace
- Recruit and test
- Observe confusion again
- Still cannot isolate feature value
Three cleaner learning loops
One audience → one job → one promise → interpretable behavior
One audience → one job → one promise → interpretable behavior
One audience → one job → one promise → interpretable behavior
At this stage, precision was more valuable than preserving every possibility.
10
FOUNDER ALIGNMENT
The debate was technology breadth versus learning clarity.
“These three capabilities form our moat.”
Keeping them together could create a uniquely powerful product.
↔
“A moat has no value if users cannot reach it.”
Focused products would let us learn what each audience actually valued with less noise and lower investment risk.
DECISION · THREE FOCUSED ENTRANCES
11
WEEK 3 · DESIGN
Strategy became three different reasons to enter.
SLIDESVisuals that take your ideas furtherFor people turning complexity into an understandable visual story.
DEMO VIDEOShow the product before explaining every detailFor founders and leaders communicating a product at a high level.
EDIT EVERYTHINGTurn a finished image back into something you controlFor creators who need to move, remove, and rewrite individual elements.
The technology stayed connected behind the scenes. The front door became specific.
12
DESIGN PROCESS
I explored broadly, then used judgment to converge.
Low fidelitySection order, promise, proof, action
→Parallel explorationFigma, Claude, Google tools, Codex, handcraft
→Founder critiqueTechnical truth and business tradeoffs
→High fidelityOne coherent direction per product
AI multiplied the options. Judgment decided what deserved to ship.
13
WORKSTATION REDESIGN
Different rooms made the user’s location visible.
IDENTICAL SURFACES → AMBIGUOUS LOCATIONDifferent layouts now tell users which room they are in.
14
NEW MENTAL MODEL
Modes organize intent. Conversion connects the rooms.
15
AESTHETIC SYSTEM
A polished output can still be wrong.
Audience fitDoes it feel credible and specific to the intended audience?
5-second comprehensionCan someone explain what it is for and what to do next?
Emotional fitDoes the tone serve the purpose and context?
Brand fidelityCould this only come from this brand?
Compositional coherenceDo hierarchy, rhythm, and tone work as one system?
Publish confidenceWould the user send it—and how much must they edit first?
The implementation remains proprietary. This case study shares the evaluation philosophy, not the private aesthetic rulebook or prompts.
16
WEEK 4 · BUILD
I carried the direction from Figma into production.
01Strategy
02Low-fi Figma
03High-fi Figma
04Figma MCP → Cursor
05Voice-prompted implementation
06Founder review + QA
07Live
17
LAUNCH
Three focused products. Three cleaner signals.
What shipped is not the end of the story.
The pivot creates a cleaner learning system: each audience now reaches one promise faster, and future behavior is easier to interpret by role and job.
18
REFLECTION
My role was to make the technology understandable enough to learn from.
I did not treat the founder’s technology as the product strategy. I turned first-time confusion into shared evidence, used field research to locate the real problem, made the tradeoff visible, and carried the resulting direction from positioning to interface to production.
Start with friction.Confusion is evidence when it is captured precisely.
Optimize the learning system.Cleaner signals can be more valuable than broader scope.
Use AI for breadth, not accountability.Judgment still owns the decision.