Self-Directed Product · AI Pet Portrait · Case Study 07
Turning probabilistic generation into a reliable commerce workflow
A generative-commerce workflow redesigned around the real customer problem: not whether AI could make an image, but whether customers trusted the likeness enough to buy it.
AI Product DesignAI AutomationGenerative AI
60→98%Measured Likeness
30→90%Strong Result Opportunity
3Parallel Previews
01 / Problem & Context
What had to become true
The challenge
A generative-commerce workflow redesigned around the real customer problem: not whether AI could make an image, but whether customers trusted the likeness enough to buy it.
My role
I owned the problem framing, system/product decisions, implementation direction, validation, and iteration represented in this case study. The emphasis is on the decisions that changed the system—not a feature inventory.
02 / Decisions & Tradeoffs
Architecture follows the constraint
Key decisions
- Use research to diagnose trust and likeness rather than assuming price or acquisition was the bottleneck.
- Convert distinguishing pet features and personality cues into structured personalization context.
- Offer three parallel previews instead of a single probabilistic output.
- Persist selected-asset identity through Stripe payment, fulfillment, recovery, and delivery.
Working loop
PROBLEM → INVESTIGATE → ARCHITECT → BUILD → VALIDATE → ITERATE
The implementation was treated as a measured loop: diagnose the actual constraint, make the smallest architectural change that resolves it, then verify behavior and quality before moving forward.
03 / System Design & Build
What I built
- Input normalization and pet-specific preprocessing feed the generation pipeline.
- Three-preview parallel generation increases customer optionality without materially increasing wait time.
- Stripe payment is tied to the selected asset and order state.
- n8n orchestrates generation, personalization, payment, fulfillment, recovery, and automated delivery as modular workflow stages.
04 / Validation & Outcomes
Evidence over claims
Validation
- Measured likeness improved from roughly 60% to 98%.
- With about 3 in 10 generations meeting the desired quality bar, three previews increased the estimated opportunity for a satisfying result from ~30% to ~90%.
- Persistent recovery paths prevent interrupted orders from requiring a full restart.
What changed
Probabilistic AI needs product architecture around uncertainty. Optionality, state continuity, recovery, and payment identity can matter as much as the generation model itself.
05 / Skills & Positioning
What this project demonstrates
AI Product Design · AI Automation · Generative AIThis case sits inside a broader portfolio spanning AI systems and agents, full-stack products and revenue infrastructure, design engineering, and creative AI leadership.
Portfolio Map
Explore the full system of work
Each project is a different proof point. The current project is highlighted so the portfolio reads as one connected capability map rather than a collection of isolated case studies.
