Home

Case study

AI Assistant

Client
Microsoft (Itaú)
Product
AI-powered digital assistant
Platforms
iOS & Android
MVP
4 months

Owned UX end-to-end for an AI-powered digital assistant for Brazil’s largest bank (98M+ customers) through a Microsoft co-innovation squad, shipping the MVP in 4 months. The assistant now resolves 1M+ inquiries per month at 85%+ accuracy.

End-to-end experience for Itaú's AI-powered digital assistant
Crafting the end-to-end experience for Itaú’s AI-powered digital assistant.

Business Outcomes

1M+
Customer inquiries resolved per month
85%+
Accuracy rate
98M
Customers at Itaú
4 months
From kickoff to MVP launch

About

Itaú, Brazil’s largest bank.

Itaú is the largest bank in Brazil and one of the largest financial institutions in Latin America with 98 million customers, 95,000 employees and 3,000 branches.

End-to-end experience overview

The problem

Customers weren’t sure they were getting the most from their bank.

Business and people problem

AI has emerged as a force capable of disrupting most value streams and business models. Customers were unsure if they were taking full advantage of their benefits / banking relationship with Itaú.

Opportunity

Introduce an AI-powered digital assistant that can answer basic questions and help customers take full advantage of their rewards.

Constraints

Four months, an agile transformation, and a language barrier.

  • We had 4 months to launch our MVP experience before code freeze.
  • Itaú was understaffed and kicking off an agile transformation.
  • Language barrier: majority of employees only spoke Portuguese.

My role

Leading design from concept to execution.

01

Identified design trends and shared AI best-practices.

02

Designed leading-edge concepts and crafted end-to-end flows.

03

Crafted new experiences and layouts while evolving Itaú’s visual design system.

04

Created prototypes in Figma to support user research and the co-innovation squad.

05

Articulated goals and concepts while presenting work to leadership and stakeholders.

06

Gave and solicited feedback from other designers and the broader product team to continually raise our bar for quality.

07

Drove alignment on product direction with cross-functional partners.

08

Facilitated workshops with stakeholders across the organization.

09

Ensured our products and features launched at the highest level of craft and execution.

How I led it

I was the only product designer from Microsoft on the project; Itaú had two product designers. Together we served eight product teams and multiple executive stakeholders, and keeping them aligned on the direction was a central part of the work.

The hardest call was the pace. The MVP was due in under four months, so I had to ramp up quickly and deliver designs in days, letting the development team ship incrementally in two-week sprints. That left no room to explore two or three options and test which one fit customer expectations best.

To stay grounded in the user, we used Microsoft Copilot to create the user personas, an industry best practice that let us move fast without losing sight of who we were designing for.

The work

From first concepts to shipped screens.

Inteligência Itaú
Building a trust layer from the start.
Initial release of the AI assistant
The initial release includes a welcome screen, AI response rating feature and next steps.
Entry points for the AI assistant
The initial release includes several entry points.
Quick security check before the conversation
A quick security check ensures the conversation is private and secure.
Welcome screen for the AI assistant
The initial release featured a new welcome screen.
AI response with rating feature and next step
The initial release featured an AI response, rating feature and next step.
AI response rating feature
The initial release featured an AI rating feature.
Designing, prototyping, and testing new features
Designing, prototyping, and testing new features.
End-to-end experience across iOS and Android
Crafting the end-to-end experience across iOS & Android.
Final designs for the talk-to-type feature
Final designs for the talk-to-type feature.
Final designs for recurring subscriptions and spending insights
Final designs for spending insights.
Final designs for recurring subscriptions and spending insights
Final designs for recurring subscriptions.
Persona for the AI assistant
Building empathy with our customers by understanding their pain points, needs and wants.
Customer journey for the personal assistant
Defining and evangelizing how an experience should behave based on people’s needs.

Deliverables

Best-practices for AI-powered assistantsBacklog of new featuresCopy recommendationsCompetitor / market researchInteraction and visual designsJourney mapsPersonasProduct roadmapService blueprintTest cases & QA supportUser flows

Outcome

Millions of customers getting answers in real time.

The AI-powered digital assistant recently launched.

By establishing an Agile approach, the team will quickly release additional features to support top inquiries on checking accounts, usage of card benefits and analysis of their spending.

Next project

Cigna, Designing for health & wellness

Led the redesign of Cigna.com across its 5 pillars and the native mobile experience, iOS and Android.

→