Viktoriia Shakhnir
Overview
AI Design Challenge is a competition for designers: pick a complex B2B product, find a real UX problem, and solve it with an interactive AI prototype.
The goal was to add an AI assistant to the analytics dashboard, making financial data accessible to beginners while speeding up work for experienced users. No constant tab-switching.
My Scope
Full cycle, solo: platform selection, sandbox research, persona building, UX problem analysis, solution design, and building an interactive prototype in Lovable.
- 1
Platform selection
Chose Chargebee: a real B2B product with sandbox mode that allows walking through the full user journey
- 2
Sandbox research
Explored the platform through sandbox with test data: from product creation to analytics review
- 3
Building personas
Defined three user types with their goals and pain points to better understand the problem
- 4
Problem analysis
Identified core UX problems: data without context, scattered analytics, terminology barriers
- 5
Prototype in Lovable
Designed the AI assistant and built an interactive dashboard prototype with a live AI chat
Research
Platform selection
I chose Chargebee - a platform for managing subscriptions, billing, invoices, and financial analytics.
Why this one:
- A real product with broad, deep functionality
- Built for B2B SaaS context, which is an area I'm interested in
- Has sandbox mode that allows walking the full user journey - from product creation to analytics review - using test data
Who uses it:
- Product managers setting up plans and discounts
- Financial analysts tracking cash flow and invoices
- Managers handling subscriptions and payment monitoring
Sandbox research
Before looking for a problem, I walked the real user path. Covered the full basic flow: dashboard overview - subscription creation - user analysis after test payments - analytics review.
User personas
To understand the audience, I built three personas: Subscription Manager, Financial Analyst, and Junior Subscription Manager.
- Need for quick access to data
- Need for automation of repetitive processes
- Need for clear and intuitive data visualization
Key findings
- Dashboards look data-rich, but they're a collection of generic metrics and charts with no explanations or next steps
- Detailed analytics are spread across separate sections in the left menu: Reports Explorer, Customer Insights, RevenueStory
- Users see lots of numbers (MRR, Subscriptions, Billing) but don't understand what matters or how it affects the business
- Getting the full picture requires constant tab-switching and familiarity with financial terminology
Challenges
Data without meaning
Chargebee shows a lot of numbers but doesn't explain what they mean or what to do next. Beginners get lost, experienced users waste time searching.
Solution The AI assistant automatically analyzes metrics and surfaces key anomalies. If there are overdue invoices, it flags them and suggests an action: check status or trigger an email reminder.
Analytics scattered across sections
Most key metrics (MRR, Retention, GRR) are only available through Reports Explorer or Customer Insights. Getting the full picture requires knowing where to look.
Solution The redesigned dashboard brings key metrics together in one place. The AI assistant is right there too - answering questions without navigating away.
Terminology is a barrier for beginners
The platform uses specific financial and SaaS terminology: MRR, Churn Rate, Net Dollar Expansion. For new users, this is a real obstacle.
Solution The assistant explains any term on request. No need to Google or open documentation - the answer is right inside the dashboard.
AI in Process
AI was part of the process on two levels: as a research tool and as a build tool.
ChatGPT helped break down FinTech terminology, shape the user personas, and think through usage scenarios for different roles.
Lovable was used to build the interactive prototype. Built an analytics dashboard with key metrics (MRR, invoices, subscriptions) and an AI chat that flags issues, surfaces analytics, and answers user questions.
Before / After
Outcome
The project took first place and a $300 prize. The mentors noted that it solves a real problem: fast, understandable analysis of financial metrics in a complex B2B product. They also highlighted the systematic approach - market research, best practices, and scenarios designed for different user roles.
For me, this project showed that AI in a product isn't a feature for the sake of it. It needs a clear context: who uses it, when, and why.
