Model the rules
I map roles, permissions, states, dependencies and failure paths before committing to screens.
Before the interface
Debora MoratallaI design AI agents, workflows and design systems people can trust.
AI guide
Short on time? Ask about me. Straight answers, with links to the proof.Selected work
01Seamline
An agent that keeps a design system in sync, and knows when to stop.
6/6breaks caught on a blind test, 0 false alarms
Solo · designed and built an AI agent
Board
A budgeting product where status had become a coordination problem.
€1.2Bbudget process, adopted by SUEZ for FY2026
Solo designer · enterprise B2B
TAP Design System
Keeping product, design and code in sync as the system grows.
55 → 38component concepts · docs from 45 min to under 2
Lead · design system and AI workflow
Fluxy
An autonomous commuting agent that knows when to act — and when to wait.
12commuting scenarios tested for trust
Lead · agentic UX concept
TAP Mindset
Turning one feature-led product into three role-based experiences.
−38%decisions before a first session
Lead · 12-month product redesign
Civeo
A municipal operations surface that preserves context from overview to action.
15internal users moved off Excel
Product designer · built in code
How I work
02
Expose the trade-offs
I make defaults, exceptions and product judgment visible while they are still inexpensive to change.
Before the debt
Stay through implementation
I bring engineering constraints into the work early and stay close enough to protect the intent as it ships.
Before the handoff// A NOTE ON IMPOSSIBLE OBJECTSComplex systems rarely look impossible all at once. The contradiction appears when you follow the rules from one decision to the next. My job is to find that break before it ships.
Make the logic visible. Keep the intent buildable.
Lab
03
Beyond my own work
04How seniors design with AI
Guest speaker at Shift+R on seniority in the age of AI, and on how I built a design system with AI and what went wrong along the way.
Workshops at Board
I run UX/UI practice workshops for product and engineering, so good decisions don't depend on one designer being in the room.
AI in the design process, with a person in charge
At TAP I set up a weekly design and engineering review: automation prepares the work, a person approves anything that touches foundations, the API or accessibility.