Web AppAI-Enhanced
Bundance
AI-powered shopping and price tracking application
{ CASE STUDY }
Signal
Real-time earnings intelligence for everyday investors - designed and shipped entirely with AI at the Figma Makeathon 2026.
Web AppAI-EnhancedFintech
Most financial tools are built for analysts. Signal is built for everyone else.
{ RESEARCH }
Why an Earnings Calendar?
I didn't start with the stock market. Using Claude as a research partner, I brainstormed across industries - healthcare, education, logistics, finance - filtering for spaces where AI could solve a real information problem, not just add a chatbot to an existing product. I narrowed by three criteria: data richness, user pain, and feasibility within the Makeathon timeline.
Starting question
Where can AI solve a real information problem - not just add a chatbot?
▼
6 industries explored
Healthcare
38.6% CAGR
Education
Fragmented market
Logistics
Slow adoption
Real Estate
Slow digital adoption
Finance
19.6% of AI market
Retail
3.7× avg ROI
▼
Finance - sub-areas evaluated
Portfolio TrackingCrowded: Robinhood, Webull, M1
Crypto AnalysisVolatile, low trust post-FTX
Tax OptimizationRegulatory, can't ship fast
Earnings SeasonScattered data, no AI layer
News SentimentBloomberg + Reuters dominate
{ COMPETITIVE LANDSCAPE }
Three categories, one gap
The incumbents
Yahoo Finance, Nasdaq, Investing.com - earnings calendars buried inside massive portals. A data table: ticker, date, EPS estimate, actual. No context, no narrative.
The dedicated apps
Earnings Hub, Stock Events - better focus, but still assume financial literacy. Built for people who already understand EPS, revenue surprises, and analyst consensus.
The AI tools
AlphaSense, Fiscal.ai, Aiera — powerful AI on earnings transcripts. But enterprise-priced ($29–$79/month+), designed for professional analysts. Inaccessible by design.
{ THE GAP }
No product takes live earnings data and translates it into plain language for everyday investors. No product answers: “What should I know about this company before they report tomorrow?” - in a way that anyone can understand.
{ PROCESS }
Breaking Traditional Design Thinking
Traditional design thinking follows a linear path: Empathize → Define → Ideate → Prototype → Test. It's a proven framework, but it assumes time that a hackathon doesn't give you. More importantly, it separates research from building - you finish thinking before you start making. My process for Signal collapsed that sequence. Research, design, and building happened in parallel, with AI accelerating every stage.
Phase 1
Research with AI
I used Claude to brainstorm industries, identify opportunities, and map the competitive landscape. This wasn’t passive research — it was an active conversation where I challenged assumptions, narrowed scope, and validated ideas in real time. Once the earnings calendar concept emerged, I verified Claude’s findings against real data: checking Finnhub’s API documentation, reviewing competitor apps, and confirming the information gap existed.
I also used my AI design twin — a version of Claude I’ve been training with my design philosophy and decision-making process — to stay true to my instincts throughout. The twin helped me evaluate layout decisions, critique my own work, and maintain design consistency at the speed the Makeathon demanded.
Phase 2
Build in Figma Make
With the concept validated, I moved directly into Figma Make. No separate wireframes, no static mockups handed off to a developer. Figma Make allowed me to design and build simultaneously — crafting the interface while wiring up real functionality. This is where the traditional process would have me in a wireframe phase; instead, I was already shipping.
Phase 3
Test with Real Users
I tested Signal with traders to evaluate whether the interface was intuitive, whether the AI-generated insights were useful, and whether the information hierarchy matched how they actually think during earnings season. Their feedback directly shaped refinements.
{ ARCHITECTURE }
How Signal Works
Two APIs working in tandem - one for raw data, one for meaning.
CLAUDE API
AI-generated intelligence
For every reporting company, Claude generates a plain-language summary: what the company does, what analysts expect, what to watch for, and how results compare to forecasts. This is Signal's core differentiator - turning raw financial data into something anyone can read without a finance degree.
FINNHUB API
Live earnings data
Company names, reporting dates, EPS estimates, actual results, stock quotes, and company profiles - pulled via a single date-range query instead of hundreds of individual API calls. Scoped to Q1 earnings for US companies: a constraint that made the product faster and more focused.
{ CHALLENGE }
Getting Live Data In
Integrating APIs inside Figma Make meant testing multiple data providers to find one that was reliable, had the right data structure, and worked within the platform's constraints. I tested several financial data APIs before landing on Finnhub.
Narrowing from all earnings, all year, to Q1 US companies wasn't a compromise - it was a design decision. The constraint made Signal more focused, faster, and more useful.
{ THE RESULT }
Signal
Signal is a real-time earnings calendar that tells you who's reporting, what's expected, and whether they beat - with AI-generated context on every company card.
The app received positive feedback from the community. It didn't win the Makeathon, but the response confirmed something more valuable: the concept has real potential. The gap between dense financial data and accessible, AI-powered earnings intelligence is genuine and underserved.
I'm considering taking Signal forward - beyond the Makeathon prototype - to build it into a more complete product.
{ LEARNINGS }
What I Learned
AI as a design collaborator changes the process, not just the speed. Research, design critique, and product functionality - AI was embedded in every layer, not bolted on as a tool.
Constraints improve products. Narrowing from full-year global earnings to Q1 US companies made Signal more focused, faster, and more useful. The limitation pushed a better design decision.
Plain language is a design choice. The AI insights on each card aren't a technical feature - they're a statement about who this product is for. Choosing to explain earnings in human language is a decision about accessibility and audience.
Figma Make enables a new workflow. Designing and shipping in the same tool eliminates the handoff gap. For a hackathon, this is transformative. For the industry, it signals where design tooling is heading.
