Available for new projects · Amsterdam

Product designer crafting design systems & DesignOps at scale.

Scalable, accessible, human-centered frameworks that help teams ship better, faster, and more consistent digital products. 7+ years at Itaú Unibanco & Loft across 22+ brands in Latin America.

R$9.7M

Saved

22+

Brands

70+

Components

View selected work → About me

Design Tokens·Multi-Brand Systems·WCAG Accessibility·Contribution Models·DesignOps·Documentation·Design Tokens·Multi-Brand Systems·WCAG Accessibility·Contribution Models·DesignOps·Documentation·

Featured projects

See all (10) →

[ Itaú Unibanco ] 02

Itaú Design Language

A white-label system enabling consistency and scale across 22+ brands.

Design System White Label Accessibility

[ Loft ] 03

Copan: Design Strategy

Loft’s WCAG-compliant system, R$9.7M+ saved, 85% adoption.

Design System Strategy Digital Design

[ Loft ] 04

Copan: Component Method

A 3-stage method for 72+ accessible components, 60% faster delivery.

Design System Web Development Brand Identity

[ Itaú Unibanco ] 05

DesignOps: Sketch → Figma

Migrating ~450 designers with research, training & a structured rollout.

Design Operations Change Management Workshop Facilitation

[ Itaú Unibanco ] 01

PDI: Developing People at Scale

An IDP program guiding the growth of around 400 people.

People Development Design Ops Career Development

[ Concept ] 06

Uniqlo DWS: Intelligent Replenishment

A speculative redesign of Uniqlo’s in-store ops app around AI-assisted replenishment.

Concept AI Product Retail Ops

View all projects →

Trusted by teams at

Itaú UnibancoLoftRede+ 22 brands across LATAM

What colleagues say

“I worked with Dri for nearly four years on the Design System and Ops team at Latin America’s largest bank. She supported over 400 designers with deep technical knowledge, empathy, and clarity. Together, we led a high-impact tool migration and managed component libraries, documentation, and access. Dri is the perfect balance between human-centered thinking and technical excellence, a rare and valuable asset to any Ops team.”

Vinicius Perussi

Senior Design · Itaú

“I highly recommend Adriana for her product design expertise, especially in design systems. At Loft, she collaborated closely with our dev team and stood out for her honesty, clarity, and constructive feedback. Her transparent communication helped us improve processes and deliver better outcomes. A pleasure to work with!”

André Seiji

Software Engineer · Frontend, Design Systems · Loft

“Adriana is one of the most skilled and reliable professionals I’ve worked with. She combines strong design fundamentals, tool mastery, and high-quality deliveries with a positive, collaborative attitude. She’s generous with her knowledge, open to feedback, and a true team player. Any team would be lucky to have her.”

Marcelo Meinberg

Squad Leader / Product Manager · Loft

Writing

All articles →

5 min read · Jul 1, 2025

UX Is Folding Shirts Too: What a Year Behind the Counter Taught Me About Systems and Care.

Read article →

← All work

Concept AI Product Retail Ops

Uniqlo DWS: Intelligent Replenishment

A conceptual redesign of Uniqlo’s internal DWS app, proposing an AI-assisted replenishment and planning system to reduce lost sales and free supervisors from manual, low-value work. Based on real operational challenges observed while working in-store, not on internal company data or an official initiative.

Open interactive prototype ↗

Client

Personal project

Industry

Retail / Fashion

Role

Product Designer (concept)

Duration

1–3 weeks

At a glance

The challenge

Manual, reactive workflows in Uniqlo’s internal DWS app left staff without real-time priorities and supervisors buried in low-value planning and troubleshooting, contributing to lost sales on best-selling items.

The outcome

A concept redesign proposing three connected systems (AI-assisted replenishment, automated scheduling, and centralized alerts) reframing DWS from a passive record-keeping tool into a proactive decision-support system.

Open interactive prototype →

Turning stock data into decisions

Manager home showing sales and an AI-suggested reallocation

Concept redesign

Empowering supervisors, not replacing them

The Intelligent Replenishment System drafts both the task and the best assignee from live sales and stock signals — but nothing moves until a person approves it.

−2h (target)

of manual planning saved, design goal not measured

Mon

Tue

Wed

Thu

Fri

Sat

Sun

Our technology

DWS reframed from passive record-keeping into a proactive decision layer.

AI Suggestions card: drafted from live sales and stock signals, matched to who's on shift

Core pillars

Driven alerts

Concept

Speculative redesign

Overview

During a year working in a Uniqlo store, I used DWS, the internal system for stock management and task scheduling, every day. It worked well as a system of record: inventory counts, task lists, and customer feedback all lived there. But none of it translated into a next action. A count of “150 units” said nothing about whether the shelf was actually empty, and task lists didn’t change when the floor did.

This concept started from that gap, not from a formal research engagement or access to internal data, but from direct, first-hand observation of daily store operations, combined with product thinking about what a system like DWS could reasonably do with the data it already had. It is a speculative redesign, unaffiliated with Uniqlo. The interface concepts shown here were designed and prototyped with Claude.

01 / Challenge

The core challenge was a reactive, manual system that created problems on two levels: on the floor, and in the back office.

◆ Inventory shown without context: “150 units in the system,” with no signal that the shelf was already empty

◆ Static task lists that didn’t reflect real-time store priorities

◆ Supervisors spending hours a day building schedules by hand

◆ Supervisors patrolling the floor to catch problems the system should have surfaced on its own

◆ Task and schedule changes communicated informally, with no confirmation or context

02 / Solution

The proposal centers on three connected systems, reframing DWS from a passive record-keeping tool into a proactive decision-support system.

Intelligent Replenishment System (IRS)

An AI-assisted layer that reads real-time sales and inventory signals to generate and prioritize restocking tasks automatically, turning “150 units in the system” into “Shelf empty, restock 15 units now.”

Intelligent Planning Module

A supervisor tool designed with a target of automating around 80% of schedule creation from sales patterns and staffing rules, so review and approval take minutes instead of hours. This is a design target for the concept, not a measured result.

Centralized Communication

A high-priority notification layer that surfaces task and schedule changes with confirmation and context, closing the gap left by informal, undocumented communication.

The Intelligent Replenishment System drafts both the task and the best assignee from live sales and stock signals, but nothing moves until a person approves it.

The Planning Module auto-drafts next week's schedule from historical sales and staffing patterns, turning hours of manual rostering into a quick review.

Centralized alerts surface stock and schedule issues in one place, each with a direct action to resolve it.

Challenges along the way

Because this is a concept rather than an implemented product, the open question isn’t technical feasibility, it’s trust. Would staff act on an AI-generated task without double-checking the shelf themselves? The design leans on transparency, showing why a task was generated, not just what to do, as the main lever for adoption.

03 / Key learnings

Improving employee tools has outsized impact

B2E changes ripple directly into customer experience and revenue. Internal efficiency is the foundation of external excellence.

Data without context is noise

The shift from “150 units” to “Shelf empty, restock 15 now” is where design turns a data point into a decision.

Automate the repeatable, protect the human

The goal isn’t replacing the supervisor, it’s freeing their time from repetitive tasks so they can focus on training and the sales floor.

Observation has limits

Grounded in real, lived problems, but without access to DWS data or user testing with other staff, it remains a hypothesis. The natural next step would be validating the IRS logic and the planning module with actual users.

04 / Result

Scale

Two user groups addressed directly: sales-floor staff and store supervisors

Three integrated systems: replenishment, planning, and communication

Grounded in real day-to-day retail operations, not a hypothetical brief

Mobile-first, sized for phone up to tablet, where store work actually happens

Intended outcomes

Faster response to low stock on best-sellers, where delays become lost sales

Supervisors reviewing and approving a schedule instead of building one

Clear, prioritized tasks replacing static, context-free lists

Fewer miscommunications around last-minute task or schedule changes

The same system from the staff side: shift progress, today's tasks, and a one-tap clock-out. The second of two audiences.

DWS taught me that the highest-leverage design isn’t always customer-facing. A companion piece to “UX Is Folding Shirts Too,” this concept comes from the same year behind the counter, arguing that the tools we hand employees quietly shape everything a customer feels.

Next project

Itaú Design Language

→