Billie
Finance.

Competitive audit of Mint, YNAB, Copilot, and Revolut revealed all four frame finance as deficit management. User sentiment patterns across App Store reviews surfaced guilt and anxiety as dominant themes — confirming a motivational gap, not an informational one.

Defined the core problem as: users know they overspend — they lack a mechanism that makes improvement feel rewarding and visible. The goal shifted from 'help users understand their finances' to 'make credit health feel like a streak worth maintaining.'

Explored three directions: solo gamified challenges (low retention without social proof), a finance social network (privacy concerns at scale), and embedded peer groups within a card management app. The third path combined utility with accountability.

Converged on Billie: a mobile card wallet with Cartels for social competition, a Gemini AI coach (the Vault) for contextual education, and a gamified XP system that rewards on-time payments and low utilisation — not just app opens.

Competitive Audit

Systematic review of Mint, YNAB, Copilot, and Revolut across six dimensions: onboarding friction, emotional tone, social features, retention hooks, notification strategy, and error framing. Identified the universal absence of peer accountability as the structural gap.

Sentiment Analysis

Analysis of App Store reviews across leading fintech apps for recurring emotional language. Guilt, anxiety, and feeling overwhelmed emerged as dominant themes in negative reviews — validating the hypothesis that emotional design, not missing features, drives abandonment.

Concept Testing

Three interaction prototypes tested across the solo gamified model, a finance social feed, and the Cartel group model. The Cartel structure showed the strongest stated intent to return and lowest reported financial anxiety — validating the social accountability mechanic.

All primary action buttons span the full container width. Icon-only controls (Send, Freeze, More) use w-14 h-14 — 56px — exceeding WCAG 2.5.3. The card eject trigger is a 56px-wide full-height button at the bottom of the wallet, impossible to miss.

Primary text on zinc-950 background exceeds 15:1 contrast. All status states (positive, negative, warning, neutral) use icon plus colour plus text label. The app is dark-first with no light mode toggle — eliminating contrast inconsistency at the cost of system preference support.

The Gemini prompt is explicitly constrained to 'supportive coaching tips, not roasts.' Mission XP rewards any improvement, not only perfection. Transaction category labels are neutral descriptors — Dining, Entertainment, Travel — without moral judgement attached.

All icon-only interactives carry descriptive aria-label attributes. The full flow uses semantic <button> elements throughout — no div-based click handlers in the entire codebase. A specification gap exists for the Vault answer panel: aria-live='polite' is the planned fix for the next iteration.

Every interaction is reachable and activatable via Tab and Enter alone. Focus order follows visual reading sequence top-to-bottom. The Ask Vault input fires on Enter key — keyboard users can complete the full AI coaching flow without touching the mouse.

Navigation labels are plain nouns — Vault, Cartel, Spendings. The Vault section provides short 'Decode' articles on credit concepts. The Gemini prompt instructs the model to 'answer clearly and simply (ELI5)', targeting accessibility for users without financial education backgrounds.

Geolocation denial shows a clear 'permission required' state rather than an empty offers list. Gemini API failures preserve the previous intel cards — the Vault never renders a blank state. All async operations are wrapped in try/catch with non-destructive error handling.

Billie makes no background API calls — Gemini is triggered only on explicit user actions. DiceBear avatars are fetched on demand per member view. The app loads from a single bundled file with no lazy-loaded chunks, making it resilient to slow connections.

Design Principle

"Design for the most financially anxious user first. If the interface adds shame, it will drive avoidance. If the improvement doesn't feel winnable, users will never return."

Traditional finance apps optimise for users who are already disciplined. Billie optimises for users who want to be. The gamification layer, the AI encouragement constraint, and the social accountability mechanic all serve this one design principle — make improvement feel possible for someone who has been failing so far.