About
I'm Carolina, a Lead Innovation Consultant at SAP AppHaus in Palo Alto, where I lead generative AI co-innovation projects with SAP customers.
I created SAP's Generative AI Explore Workshop format and have run it with leaders from Google, Apple, Microsoft, Intel and Petrobras, training more than 100 people along the way.
I trained in industrial design in Rio de Janeiro and in human-centered design at the Bauhaus in Dessau. For more than fifteen years I've been turning research into products people actually use, from public-sector data tools to enterprise AI.
Practice
- UX strategy & design leadership
- Human-centered design
- Design thinking workshops
- User research
- Prototyping
Focus
- Generative AI & AI agents
- Enterprise software
- Service design
- Data visualization
Tools
- Figma, SAP Build
- SAP Fiori design system
- Mural, Jira
- Claude Code
Selected work
3 case studiesAI Supply Chain Troubleshooter
The challenge
How might we use generative AI to find the causes of sales order commit issues and take corrective action, so specialists work faster and customers stay satisfied?
When a sales order drifted from the ship date promised to the customer, AMD's supply chain specialists had to find out why. Each request meant working through 14 manual checks in SAP S/4HANA, and the team received hundreds of these requests every week.
What we found
Discovery sessions with specialists pointed to two root problems: no contextualized insights, since the system gave no reason when an order wasn't committed, and time-consuming manual checks across allocation, supply, order and availability data.
The solution
A conversational troubleshooter built on SAP BTP. A specialist types a sales order number in plain language and the assistant runs the checks, shows each result with its reasoning, and answers follow-up questions. Results can be turned into a customer email.
"We must uncover the root cause of why sales orders were not committed as expected. The system behaves like a black box, leaving us in the unknown."Supply chain specialist, discovery research
"We receive hundreds of weekly requests from over a hundred users, but due to bandwidth limitations, we lack the manpower to handle them all effectively."Supply chain specialist, discovery research
Explore
- Identified 3 high-value use cases
- Picked the front-runner
- AMD and SAP executive sponsorship
Discover
- End-user discovery sessions
- Persona
- As-is journey and pain points
- To-be journey
Design
- Design prototype
- Technical prototype
- Feedback sessions with business and IT
Deliver
- Productive pilot
Run & Scale
- Scaled solution with additional request types
Allocation
- Wildcard record
- Available allocation is zero
- Negative allocation
- Misalignment on order
- CAPA constraining
Supply
- No or limited supply
- ABC disabled on order
- Supply at excluded area
- Order at excluded location
- Segment check
Order
- Header-level delivery block
- Partial delivery prevents ABC
- Fixed date/quantity enabled
Availability
- ATP not applied
Open clickable prototype →Less time and cost spent on root cause analysis
More efficient process, from 20 to 2 minutes per request
Estimated time saved, at 100 requests a week
A high-flying invoice assistant powered by AI
The challenge
Cirque du Soleil processes 70,000 invoices a year, and about 40% of the inquiries reaching Accounts Payable ask the same thing: "What is the status of my invoice?"
Each inquiry meant an AP specialist had to search for the invoice by hand and write a reply. Inquiries arrived in French, English and many other languages, so specialists also needed translation to understand each request and to reply in French and English.
The process
We ran Human-Centered Design AI workshops with the AP team to find which parts of their work AI could take on: reviewing emails, translating, investigating invoice status and drafting replies. The specialist stays in control and reviews every response before it goes out.
The outcome
Urgent and sensitive emails are now identified and handled within the hour, skipping the usual 48-hour delay for spotting urgency, escalating and processing. Other requests get answered more than 25% faster, and the AP team's backlog dropped by 25%.
- TranslateInto English or French, the team's working languages
- PrioritizeCategory, summary, urgency and sentiment
- ExtractInvoice numbers, even from attachments and tables
- Look upInvoice and payment status in SAP S/4HANA
- DraftA reply informed by similar past answers
- Translate backInto the sender's own language
Open clickable prototype →Improvement in handling urgent or sensitive requests
Faster responses to non-priority requests
Smaller request backlog for AP specialists
"GenAI makes us more efficient and consistent. It helps us focus on our customers, gives us more complete responses to invoice emails, and cuts down on back-and-forth with clients."Team Lead, Accounts Payable
Course Advisor
The challenge
How might we help the school's Program Advisors see each student as a whole, more efficiently, and make intentional course recommendations based on insights drawn from the school's data?
The process
Following SAP's Human-Centered Approach to Innovation, I began with the advisors themselves: how they gathered student information today and what they needed to recommend courses with confidence.
The concept was designed on SAP Business Technology Platform, bringing student information together in one place and pairing it with intelligent insights.
Discovery artifacts
The research behind the concept: who we spoke to, what we learned, and the principles that shaped the design. Use the arrows or your keyboard to move through the slides.
Understanding advisors before designing for them
Before any screens, we spent the Discovery phase with the people who guide MBA students through course choices, to learn how advising works today and where it breaks down.
of MBA students don't know which courses to take when exploring new career pathways.
- Students trust advice from peers more, because their experience feels relatable.
- Advisors saw data-driven insights as a way to feel more confident in their advice.
- The school's goal: a white-glove advising service that brings students closer to their professional goals.
We interviewed the whole advising ecosystem
Mapping who is involved, then planning the work
Three kinds of students walk into an advisor's office
Path Followers
Know which courses to take and have clear career goals. They stick to the tried and true.Path Adapters
Think they know what to take but aren't sure, and are open to other options.Path Explorers
Don't know what to take. Often career changers heading off the beaten path."Out of 10, 2 students know what to do."Program Advisor
Kim Spring
Program Advisor
An experienced advisor with a project management background and a very busy year: meetings with students, email follow-ups, and events.
Helping students move through the program successfully and reach their goals.
"We are very good on the personal and support side, but on academic advice we are sometimes weak. It is hard to know all the course details to advise students appropriately."Program Advisor
Pain points concentrate in course search & selection
Students search four or more places (faculty website, class catalog, Class Choose ratings, Google) and still can't find what fits their goals.
Advisors aren't trained in every industry, which lowers their confidence in recommending specific classes.
Both struggle to match a huge variety of classes to a student's interests and goals.
What we learned about students and advisors
- Trust guidance that is evidence-based: peers' experience with a class, and where graduates ended up.
- Want to build their network through the school's ecosystem.
- Often change goals: a first-year set on consulting may pivot a year later.
- Need data-driven insights: which skills a course builds and where past students went next.
- Want to know the rationale behind any recommendation.
- Need one holistic view of each student instead of scattered sources.
Two problems, two opportunity areas
How might we help advisors understand the "student as a whole" more efficiently?
Connect each student's goal setting to course recommendations, and give advisors access to those goals from the first year until graduation.
How might we give advisors intentional course recommendations that meet their expectations?
Simplify access to useful recommendations and give advisors a single source of truth.
"Being a student at [the school] is like drinking from a fire hose. There's an abundance of opportunities."Program Advisor
Four principles to design by
Flexibility
Allow for changes along the way. Students' goals shift, so recommendations must adapt.Transparency
Explain the why behind every recommendation, not just the what.Intentionality
Begin with the end in mind: start from the goals and skills the student wants.Inclusivity
Bias-free results for every student profile. Secondary research showed algorithms at other universities treating students unequally."Make clear what the ML is recommending. Based on what factors of personalization?"Program Advisor
The reframed design challenge
How might we help the school's Program Advisors see each student as a whole, more efficiently, and make intentional course recommendations based on insights drawn from the school's data?
Understand why a class is recommended
See a roadmap from where a student starts to where they want to end up
A high-fidelity clickable prototype for the school's student advisors
The outcome
The original prototype used machine learning: advisors tuned results with a conservative-to-experimental slider and attribute boosts, and explored course clusters built from historical data. I reimagined it with generative AI. The advisor gets a plain-language brief of each student, asks for recommendations that explain themselves in terms of the student's own goals, refines them in conversation, and prepares the meeting and follow-up in the same place.
Open clickable prototype →- National Hockey LeagueVenue Metrics. An end-to-end cloud platform for venue data, featured in Forbes. Learn more ↗
- SAPVaccine Collaboration Hub. Nominated for the UX Design Awards, 2021. Learn more ↗
- German Environmental AgencyIndustry pollution transparency tool. A searchable map of pollution data from more than 5,000 industrial sites, designed for a public audience of 80 million.
- LufthansaInnovation Accelerator. Facilitated 10+ design thinking workshops and built 20+ prototypes to validate employee-led concepts.
- FGV OnlineDistance learning platform. Flows and interactions for a platform serving 2 million students.
Experience
- 2022 – nowLead Innovation ConsultantSAP Labs · Innovation Experience Services, Palo Alto. Leads GenAI co-innovation projects and a small team of designers.
- 2020 – 2022Senior User Experience DesignerSAP Labs · SAP AppHaus, Palo Alto
- 2019 – 2020Interaction DesignerDaylight Design, San Francisco
- 2017 – 2018User Experience DesignerSAP SE, Walldorf and Berlin · S/4HANA and Innovation Center Network
- 2010 – 2018Human-Centered Design ConsultantIndependent, Berlin
Education
- 2025 – 2027LEAD Leadership Program, Stanford Graduate School of Business
- MAIntegrated Design, Anhalt University of Applied Sciences, Bauhaus Dessau
- BAIndustrial Design, Federal University of Rio de Janeiro
Recognition & community
- 2022SAP Heroes, one of 50 employees recognized for contributions to local communities
- Since 2022Leader, SAP Sustainability @ Silicon Valley employee group
- 2021 – 2022Sustainability Advisory Board, San José State University
Contact
Let's talk about your next AI product.
