Try adjusting the filters or search term to find what you are looking for.

No articles found

Monthly-ish stories from the studio: experiments, design learnings, and the tools I lean on to craft better products.

Join the Creative Field Notes

Blog

I am preparing this book with Casa do Código to help designers and product teams adopt AI without outsourcing judgment. The draft focuses on practical workflow changes, source validation, ethics, and the parts of design that still need human review.

A Casa do Código book in preparation about design, artificial intelligence, and safer team workflows.

Design in the AI Era (in preparation) - Thiago Xikota

Deciding what AI should help with, and what stays human.

Creative direction

Understanding teams, constraints, and workflow risks before choosing tools.

Context mapping

Using AI to organize evidence without inventing confidence.

Research synthesis

Turning intent into testable UI, content, and process drafts.

Prompting and prototyping

Checking provenance, errors, and hallucinations before work ships.

Source validation

Defining guardrails for data, authorship, accessibility, and impact.

Ethics and safety

The outline can still change. These are the questions guiding the work.

Draft themes

A book in preparation about judgment, ethics, and AI workflows for design teams.

Design in the AI Era

I am preparing a Casa do Código book about how designers and product teams can use artificial intelligence with judgment, ethics, and safer workflows.

**Scope:** Senior designer in an ecosystem of 300+ collaborators, simultaneously leading design across 4 distinct Figma files (mobile app with 14M users, web admin system, Bold DS, and DSGov style guide) + accessibility research. I worked with 3 Product Owners, 2 engineering squads, and federal government stakeholders (Ministry of Education).

**Decision authority:** Full autonomy over design decisions and information architecture. For changes affecting multiple squads (like the navigation flowchart), I consulted PMs before implementation. My accessibility findings were presented to the federal government's DSGov team and influenced their roadmap.

**Timeline:** 3 years (Jul 2022 - Jul 2025). The biggest impacts (enrollment form, financial aid, accessibility) happened in 2-4 sprint cycles with continuous validation via Firebase Analytics.

**Impact outcome:** +25% completion rate on financial aid enrollment for 14M students. Accessibility thesis published at a govtech conference. Bold DS became an accessibility reference in the Brazilian industry.

Strategic Context

Bridge Lab (UFSC) connects government and citizens through technology. The flagship product is Student Journey: a Ministry of Education app used by 14 million students across Brazil.

The app serves as digital student ID, enrollment tracking, the Pé-de-Meia program hub, and Ministry services. The web admin lets Ministry staff manage credentialing, notifications, banners, content, and audits.

To understand the real operating context behind the product, I combined Hotjar sessions, Firebase data, Metabase dashboards, support tickets, and stakeholder interviews. I consolidated that material into three key personas: Janaína (project coordination and cross-team follow-up), Silmara (executive decision-maker under political urgency), and Rafael (young operational profile with high digital fluency). That framing made one thing obvious: the ecosystem could not work only for students. It also had to communicate status, risk, and next steps to the people sustaining the program behind the scenes.

The Context

The financial aid enrollment flow had concerning completion rates. Target audience: public school students on entry-level phones with spotty connections. The PM brought Firebase data: 35% abandoned mid-form.

**First attempt (failed):** Consolidated 6 stages into one page. Logic: fewer screens = less friction. Result: abandonment jumped 15%. Crash logs showed dropped connections caused all progress to be lost.

**Second attempt (success):** Modular stepper with 5 steps, each independently saveable. Auto-save on every field. Connection drops at step 3? Resume at step 3.

**Measurement:** Compared completion 60 days before and after (Firebase Analytics, new signups only). Completion rose from ~55% to ~75% (+20 percentage points). Scale: 2M+ students went through the flow.

The Form Problem

The most sensitive module: personal data, photos, dependents. A 14-year-old in rural Maranhão and a 17-year-old in São Paulo must complete the same flow with equal clarity.

Mapped 12 scenarios with the product owner, from happy path to edge cases affecting hundreds of thousands:

- With/without dependents (interfaces branch within the same flow)
- Missing government data (blank fields need clear visual fallback)
- 4+ line names (common in Brazil)
- Preferred name vs. full name hidden (minor privacy, legal requirement)
- Photo swap/deletion with destructive confirmation
- 6 error scenarios with recovery actions

At 14M users, what others call 'edge case' is 500,000 real people.

Digital Student Identity

Federal scholarship designed to keep students in school. I led the design of the full module: enrollment, eligibility, payment status, installments, and communication.

Complexity: conditional eligibility (attendance, enrollment, profile data), multiple simultaneous enrollments, installments with different statuses, and Ministry support integration.

**Decision I defended:** Each eligibility state gets its own screen with specific copy and clear action. The PM wanted one generic screen with dynamic messages. I argued with support data: 40% of helpdesk calls were 'I don't understand my status.' Dedicated screens reduced ambiguity.

Beyond the in-app flow, I also designed communication assets, detailed installment states, and a responsive web surface for following the program in the browser. That kept MEC and Pé-de-Meia branding visible across touchpoints without sacrificing clarity.

Support confirmed noticeable improvement after deploy.

Pé-de-Meia Program

12+ modules for Ministry staff. I convinced the tech lead to adopt one consistent pattern: list with filters, create/edit form, color-coded status indicators, breadcrumb navigation, and pagination. Managers who learn one module already understand all.

The embedded satisfaction survey (emoji scale + free text) was my proposal for collecting in-app feedback from public-sector managers. That data fed backlog prioritization.

I also designed from scratch the web entry point for the Student Journey ecosystem: login, responsive browser experience, and Pé-de-Meia tracking surfaces outside the app. The goal was to expand access without breaking DSGov compliance, MEC branding, or the product's overall coherence. When your stakeholder is the Ministry of Education, visual consistency is institutional infrastructure, not decorative polish.

Admin System and Web Experience

My Computer Science thesis (top grade, UFSC) was a comparative accessibility audit of two design systems. Not isolated academic work: findings directly impacted the product.

**Methodology:** Tested with 6 tools (Lighthouse, Axe Core, WAVE, JAWS, NVDA, VoiceOver) across 15 components from each system. Found 277 accessibility violations, from insufficient contrast to keyboard navigation bugs.

**Impact:** Presented to Bridge leadership. 23 critical violations prioritized for fixes. Government system findings reported to the federal government. Thesis became internal reference for accessibility decisions.

Bold is Bridge's open-source design system: 120+ components with usage guidance, interactive playground, and state documentation.

Accessibility: Thesis & WCAG Audit

The federal government design system is mandatory for gov-tech. I created a tailored style guide because official components had inconsistencies between documentation and implementation. Covers colors, typography scales, and customized components within government rules.

Also created the complete app navigation flowchart: Login → Onboarding → Home (8 sections). With 300+ collaborators and 4 squads in parallel, this map became the most-referenced alignment artifact; PMs and devs checked it before starting any feature.

Style Guide & Navigation Map

**Scope:** I was the sole product designer in a 35-person B2B2C SaaS (Brazilian engineering, European management). I served as Lead Product Designer + interim Product Manager, reporting directly to the CTO.

**Decision authority:** Proposed product architecture, defined feature prioritization with the squad, and led validation experiments. For structural decisions (like the dual-persona dashboard architecture), I discussed with the CEO before implementation, but presented the evidence that convinced him.

**Timeline:** 12 months (Sep 2024 - Aug 2025), working in 2-3 week cycles with continuous releases.

**Structural outcome:** Transformed 50% abandonment into a viable product model for 100+ productive clients. The shift from 'one product with two conflicting users' to 'one product with two integrated contexts' became the market case that Multidrop still uses for recruitment today.

Multidrop had 50% abandonment in onboarding. I discovered this by cross-referencing Mixpanel data with Hotjar session recordings: new users landed on empty dashboards, clicked 2-3 times with no result, and left. I interviewed 8 users who abandoned to understand why: the problem ran deeper than UI.

The product served two completely different personas. The Creator wants to build products, set up checkout, manage affiliates, and analyze revenue. The Affiliate wants to find products to promote, track commissions, and view performance. The dashboard treated both as the same user.

The CEO wanted two separate products. I convinced him that would double maintenance costs and fragment the user base. The alternative: one product with two contexts. It took 3 weeks of prototyping and 2 rounds of testing with 12 users to prove the persona toggle worked.

The Problem

The first version of the toggle was a dropdown in settings. In testing, 4 of 6 users couldn't find it. The second iteration: a toggle visible in the header, hit 100% discovery in the same tests.

The toggle completely changes: sidebar navigation, dashboard metrics, KPI cards, and available flows.

For the **Creator**: gross revenue, revenue source (direct vs. affiliate), trends over time, sales by product and affiliate.

For the **Affiliate**: total commissions, pending balance, next payout, clicks and conversions.

The engineering team resisted: 'it's like maintaining two apps.' I showed that 70% of components were shared. The frontend lead approved after reviewing the Figma prototype.

Dual Persona Architecture

First attempt: a guided tour with tooltips (industry standard). Hotjar showed 80% of users closed it at the second tooltip. The tour taught features instead of answering 'what do I do first?'

The iteration that worked: turning the dashboard into a visible progress space. The 'Getting Started' widget with a progress ring (starting at 0%) isn't a tutorial; it's a list of unlockable wins. Each step saves server-side, so unstable connections don't lose progress.

For the Creator: 4 steps. For the Affiliate: 3 steps. Same system, tuned per persona.

**Measurement:** Compared onboarding completion 30 days before and after deploy (Mixpanel, new signups only). Abandonment dropped from 50% to 25%, validated with Multidrop's data team.

Onboarding: From Failure to Results

The split-screen signup was a decision I defended against marketing, who wanted a simple form 'like everyone else.' For a European startup competing with Hotmart and Eduzz, the visual needed to say 'we're different' before the user read a word. Left: clean form, Google OAuth first. Right: dark, fluid, premium imagery.

The role selection screen (Creator vs. Affiliate) right after signup is where the system first splits. Radio cards with icons and plain-language descriptions, tested with 5 users who'd never used info-product platforms.

Signup as Positioning

Most systems treat refunds as compliance: a generic form to avoid chargebacks. I proposed to the PM that the 6 refund reasons ('Can't access yet,' 'Quality below expectations'...) be treated as product data, not just legal categories.

**For the buyer:** 3-step flow: find purchase by email, see details with deadline, select categorized reason.

**For the Creator/Multidrop:** each reason feeds a quality dashboard. 'Can't access yet' repeated 15 times for one product? Broken onboarding signal.

The PM wanted generic Stripe reasons. I showed that generic 'Other' accounted for 60% of selections, useless data. With domain-specific reasons, 'Other' dropped to 12%.

Refunds as Product Research

**Scope:** Designer at a ~100-person foodtech/agtech, working as one of two product designers on the team. Responsible for key features across two products: Rastreador (ERP + traceability) and CliCQ (quality inspection + workflow builder).

**Decision authority:** End-to-end UX/UI on both products, with discovery alongside industry operators, prioritization with PM/CEO, and continuous contribution to the existing design system. I worked in pair with the other designer to align patterns across Rastreador and CliCQ.

**Timeline:** 6 months (Jun 2025 - Nov 2025), shipping critical features across Rastreador and CliCQ while contributing to the existing DS.

**Structural outcome:** Contributed 10 new components to the existing DS, standardized handoff between the two products, and implemented a Figma + Make + MCP pipeline that reduced spec rework and accelerated dev sync.

PariPassu sits at the intersection of ag-tech and food-tech: traceability and food compliance. Two distinct products: Tracer (food chain traceability) and CliCQ (audits and compliance), each serving different personas: field operators and quality coordinators.

A DS was already in place when I joined, but was misaligned across the two products, which had grown independently for years. ~40% of sprint time went into repetitive visual specs that could have been standardized.

The Ecosystem

The design system was already in place when I joined. My work was to manage its evolution alongside the other designer, contribute 10 new components, and standardize handoff between Tracer and CliCQ.

**Collaborative adoption:** Started with high-frequency components (inputs, buttons, tables) with usage docs and "when NOT to use" guidance. Weekly pairing with each dev to align real usage and tune tokens.

**Handoff automation:** Built a Figma → Make → MCP pipeline: when a component is published, specs are extracted automatically and devs are notified in Slack with colors, sizes, variants, and a direct link. Reduced handoff friction without requiring a DS rebuild.

Design System: Ownership & Adoption

Import was the biggest friction point. Operators register hundreds of movements daily; manual entry was impossible. I interviewed 4 operators and found each had a different spreadsheet format.

**Critical decision:** Manual column mapping over auto-detect. The engineering lead wanted automation. I argued: mismatching a 'Batch' field can trigger a product recall. Accuracy beats speed when errors have regulatory consequences.

3-step flow: Upload → Column mapping (required fields marked) → Confirmation. Each step independently saveable.

Tracer: File Import

The Purchases screen shows food-chain complexity. I defended: 'Sync pending' status with different colors per ERP (TOTVS vs. Sankhya). The PM wanted a generic 'Pending' tag.

Support logs: 3 of the top 5 tickets were 'my purchase didn't sync' with no context on which system failed. Visual tags let operators identify the failing system in 2 seconds.

The product mapping modal solves another pain: 'Banana nanica - 1Kg' in Tracer vs. numeric codes in the ERP. Visual mapping with dropdowns ensures correct sync.

Tracer: Purchases & ERP Sync

The hardest challenge. Non-technical food-industry users needed to build conditional logic: 'If inspection score < 70%, alert manager AND trigger detailed inspection.'

**First attempt (failed):** Sequential form with text conditions. Tested with 3 quality managers; none completed a branching flow without help. Their mental model was visual, not textual.

**Second attempt (succeeded):** Canvas with 3 draggable block types connected by arrows. Benchmarked against eVisit and Zapier. Same 3 managers: all completed a branching flow in under 5 minutes, no help needed.

**Form Block (blue):** which form, owner, notification, deadline.
**Condition Block (red):** score ranges, classifications, per-question rules.
**Result Block (green):** approve, reject, or chain another form.

CliCQ: Visual Workflow Builder

Form builder with 7 question types, each with 3 states. Customizable inspection reports: PDF layout, photo grouping, logo, email settings per recipient language.

Mobile inspection has two modules:
- **Product:** QR scan → Form → Supplier → Confirm → Inspect
- **Process:** Select Form → Select Evaluator → Inspect

Separating them came from field observation (I visited 2 distributors with inspectors). Product flow starts with physical scan. Process flow starts with person selection. Mixing caused errors.

CliCQ: Forms & Inspections

In past enterprise product design work at Wellhub (formerly Gympass), I worked on critical flows for a global corporate wellness platform.

Due to confidentiality agreements, detailed case study content including specific metrics, internal processes, and feature details cannot be shared publicly.

For a full portfolio walkthrough with context on my contributions, please reach out directly.

About This Work

Wellhub (formerly Gympass) is a global corporate wellness platform connecting companies, employees, and wellness partners. I joined a core product squad responsible for critical user-facing flows in the platform.

The Platform

I led end-to-end design for critical user flows, from strategic framing to pixel-level handoff. My work focused on:

- Redesigning high-stakes user flows with comprehensive edge case mapping
- Championing transparent error handling over hidden failures
- Driving accessibility standards across multiple screens
- Contributing to the centralized design system
- Creating handoff frameworks adopted by other teams

Each design decision was validated through analytics and user feedback, with continuous iteration in 2-3 week cycles.

My Contributions

At this scale, every edge case is experienced by millions of people. My approach centered on:

**Transparency over simulation:** When systems fail, users deserve honest feedback. I advocated for transparent error states that build trust rather than hiding failures.

**Accessibility as design thinking:** WCAG compliance wasn't a checkbox. The constraints forced clearer hierarchies and more explicit states that improved the experience for all users.

**Data-informed iteration:** Every flow was measured before and after. Decisions were backed by support data, analytics, and user feedback.

Design Approach

While specific metrics are confidential, my work contributed to:

- Improved user experience across critical product flows
- Accessibility improvements adopted as standards by multiple teams
- Error handling patterns that increased user trust
- Design system contributions benefiting the broader product

The detailed impact and metrics are available in a live portfolio walkthrough.

Impact

This case study is intentionally brief due to confidentiality. I'm happy to walk through my process, decisions, and impact in detail during a conversation.

Reach out via LinkedIn or email for a portfolio walkthrough.

Want to Know More?

Design progress

This longform case study is currently maintained in Portuguese.

Impact beyond the screen

I design product experiences for millions of users. With 7+ years across startups, scale-ups, and companies worldwide, I'm passionate about building delightful, accessible interfaces that drive real impact.

Interested in working together?

What the research revealed

All my certificates and courses, with filters and search.

Certificates & Courses

For recruiters and interviewers: I can walk through process, decisions, and learnings in a private conversation. Reach out on LinkedIn.

Work in progress, under confidentiality

I'm always excited to discuss new projects, creative ideas, or opportunities to collaborate. Feel free to reach out!

Get in touch

All fields marked with * are required so I can reply properly.

Message sent successfully

Send a message

Share a few details and I will get back to you within two business days.

Start a conversation about your next project

Product design insights, accessibility, and AI applied to UX.

Product Notes

Privacy first

What happens next?

I will review your message and follow up within two business days. If you need a faster reply, feel free to send an email directly.

Message received with success

Let's build together

Quick chat

Recent coursework across UX research, accessibility, design systems, AI, and front-end implementation.

EF SET English Certificate (C2 Proficient)

Excel Course

Italian Language Course (Level B2)

Intensive Oratory Course

Design Track

Pixel Init - Introduction to Junior Enterprise

Power BI Course

Selective Process Course

Character Design Workshop

Selected certifications that reinforce research, product thinking, interface systems, and technical fluency.

Courses & Certificates

Completed the Italian Language Course (Level B2) at the Pontifical Lateran University in 2025, strengthening communication skills and cultural fluency during my academic period in Rome.

Pontifical Lateran University: Italian Language Course (Level B2)

Specialization in Artificial Intelligence at Sapienza University of Rome, focused on AI systems, research, and the technical foundations behind modern workflows.

Sapienza University of Rome: Specialization in Artificial Intelligence

Graduated in Computer Science from one of the best universities in Latin America, with a solid foundation in software engineering, algorithms, and human-computer interaction.

Federal University of Santa Catarina: Computer Science

Completed a year of high school in Vancouver, improving English fluency and experiencing cultural exchange through daily academic life in a Canadian environment.

High School in Vancouver, Canada

Formal training, international study, and continuous learning shaped around product, systems, and implementation.

Education

API error

Something went wrong

An unexpected error occurred while processing your request.

An error occurred - Thiago Xikota

Connection error

The page you're looking for wasn't found.

Page not found - Thiago Xikota

Page not found

Server error

Request timed out

Access denied

Error {{statusCode}}

If design is the passion, fixing links is the plot twist.

Unknown Error

Founding lead for the Florianopolis group, organizing local AI conversations, events, and practical learning for people building with artificial intelligence.

Lead: The AI Collective Florianopolis

Product Designer for Ministry of Education (MEC) solutions. Redesigned critical journeys for a national app with millions of users, utilizing the public design system (DSGov) and scaling libraries.

Product Designer: Laboratório Bridge

Led the UI track, mentoring designers and running workshops and video lessons.

Drove rebranding, partnerships, and events that aligned strategy, execution, and design governance.

UX/UI Designer: Pixel - Soluções Digitais

Lead Product Designer in a high-growth startup. Led discovery, roadmap and Design Ops, building a reusable design system and cutting handoff across 3 products and 25 features.

Lead Product Designer: Multidrop

Worked in a food traceability SaaS, structuring design systems and product processes to keep product, engineering, and business aligned.

Design Systems and Product: PariPassu

Past enterprise-scale product design work at Wellhub, the global corporate wellness platform. I use that experience to explain how large organizations create messy core problems and how AI can help teams address them with more leverage.

Enterprise-scale proof: Wellhub (formerly Gympass)

Independent studio at xikota.design helping teams adopt AI in design, product, and content workflows with training, workflow audits, and human-reviewed automation.

Founder: Xikota Design

Where I've made impact

Curious minds ask

Xikota Design

and lead of

designing systems and AI workflows for products that serve millions.

Product Designer with 7+ years in product, Computer Science training, and AI specialization at Sapienza University of Rome. Founder of

The AI Collective Florianopolis

From 0 to 100 Clients

Prompts, tools, references, and working methods I use with teams when AI needs to survive real review.

The AI design library I use in real work

AI specialist, founder of Xikota Design, and lead of The AI Collective Florianopolis. I help teams use AI safely in design, product, and operations workflows.

AI Specialist and Founder

14M Students · Accessible redesign at scale

Education Admin

Rastreador · CliCQ · ERP integrations

Food Traceability

Enterprise corporate wellness platform · past product design proof

Global Wellness Platform

Founder practice · AI workflows · design consultancy

AI, Brand, and Interface Studio

Wellhub: Global Wellness Platform

A running log of decisions that improved my product thinking after they first went wrong.

Solving for 14M Users When You Can't Change the Rules

Propose Broad, Ship Narrow

What failed, what changed, and what stuck.

Lessons Learned

You're offline

Public education product for 14 million Brazilian students.

Bridge - MEC

You are requesting access to the {{project}} case study.

Incorrect password

Too many attempts

Security verification required

Access granted

Access unavailable

Confidential case study

DSGov forced me to redesign navigation for WCAG AA. The result wasn't just accessible: it completed 25% more. Accessibility isn't compliance, it's a design tool.

Constraints force clarity

At Bridge, each Bolsa Familia state has different rules for 14M students. At Multidrop, two completely different user types shared one product with a hidden toggle that 4 of 6 testers couldn't find. I design for the scenarios most ignore.

Edge cases at scale are millions of people

At Multidrop, I built the DS from scratch to unify the experience for two very different user types in a single product. A design system isn't a library; it's a shared language that changes how the whole team makes UI decisions.

Systems outlast screens

How I think about design

Founder of Xikota Design and lead of The AI Collective Florianopolis. I teach teams how to bring AI into everyday workflows with safety, better decisions, and less noise. My product background includes enterprise-scale work at Wellhub, gov-tech at Bridge, and startup systems at Multidrop.

AI, design, and real work

Explore some of my highlight projects that showcase my design approach and problem-solving skills.

Featured Projects

Landing page for a non-profit organization that brings international students to the Vatican City to study Canon Law. The project focused on product design, UX, and UI to create an inviting, accessible, and informative digital presence.

Academia Canônica

All published projects now have a dedicated route. From home to the listing, and from the listing to each case, navigation is continuous.

Resources for designers using AI with judgment

Structural convergence in AI-generated media

Design no Figma com Claude Desktop: setup gratuito e guiado

Libraries

Vocabulário de frontend para designers

Ferramentas de IA para designers

Referências e leituras sobre design e IA

Frameworks e dicas práticas para design com IA

Biblioteca de prompts para designers