Erns

An AI-powered financial terminal that levels the playing field by turning dense SEC filings into accessible, Wall Street-grade data for everyday retail investors.

Comment

Erns - Earnings Intelligence

GenAI Devs Hackathon Submission

Social Good Category: Financial Literacy and Accessibility

The Mission: Wall Street relies heavily on expensive quantitative analysts, customized software terminals, and armies of researchers to gain an edge. Meanwhile, retail investors and everyday citizens are left trying to decipher dense and incomprehensible SEC reports on their own. We built Erns to democratize financial intelligence. By using advanced AI to process complex earnings reports like the 10-K and 10-Q as well as fundamental data, we condense everything into a clean and intuitive platform. We strongly believe that high quality financial data should be accessible to everyone and not gatekept by large institutions.

[!IMPORTANT] Important Notes for Judges:

  • Backend Initialization: Please interact with api.tychefinancials.com first. The backend is currently hosted on a free tier service on Render and it often spins down when it is left idle. Pinging the URL will wake up the server so the app can run smoothly.
  • Rebranding in Progress: The project was originally conceived as Tyche, which explains our current domain name tychefinancials.com. We have since rebranded the project to Erns and we will be officially moving to the domain ernsdata.com in the near future.
  • Demo Billing Mode: Everything you see in the application is currently a live demo, including the billing system. The application will prompt you to select a Stripe plan for the higher tiers, but all transactions and credit card prompts are mathematically fake and running in test mode so it will not actually cost you any money.

Inspiration

The inspiration for Erns was born out of frustration. As students and retail investors, we realized how heavily the deck is stacked against the average person in the financial markets. The SEC holds a massive amount of company facts in XBRL format, but this data is notoriously difficult to parse and synthesize for someone without a technical background. We wanted to build a modern and lightning fast dashboard that rivaled expensive enterprise financial terminals. Our goal was to build this tool exclusively for everyday people to empower them with complete data transparency and help them make better financial decisions.

What it does

Erns is a comprehensive Earnings Intelligence platform. The system functions as a high performance terminal that dynamically aggregates fragmented data sources from across the web. The platform pulls live company facts and filings natively from the SEC database. It also integrates market data from Polygon and Yahoo Finance. From there, it aggregates earnings calendars, tracks company guidance, monitors market signals and anomalies, and offers a robust stock screener. Finally, it presents all of this information in a beautiful, responsive, and data dense web dashboard built with Next.js.

How we built it using AI

Because this is a Generative AI hackathon, we built this platform entirely with an AI native development workflow. We relied heavily on ChatGPT for initial brainstorming and architectural planning. Once we nailed down the vision, we used Antigravity and Claude as our primary coding assistants.

These tools allowed us to rapidly architect our complex backend data pipelines and helped us precisely structure our massive 5000 line monolithic backend file. We also used them to instantaneously generate robust Next.js UI components and perfectly align our frontend designs without getting bogged down in boilerplate code.

The Tech Stack:

  • Backend: We used Python and FastAPI with an SQLite database and custom scraping architectures. We deployed the backend using Render for simple and scalable hosting.
  • Frontend: We built the web application using Next.js 16, React 19, TailwindCSS 4, Shadcn for UI components, and Framer Motion for animations. We also integrated Clerk for user authentication and Stripe for payment processing. We used Vercel for our frontend hosting to ensure lighting fast global delivery.

Challenges we ran into

Building a unified API to normalize fragmented financial data was an incredibly difficult process. The SEC API specifically has very strict rate limits and variable data structures for different companies, which made building a reliable parser challenging. Additionally, managing the aggressive rendering demands on the frontend was tough. We had to ensure that massive JSON payloads containing years of financial history did not lock up the user browser. This required us to implement significant optimization and caching strategies between our FastAPI backend and the Next.js frontend to maintain a smooth user experience.

Accomplishments that we are proud of

We are immensely proud of building a platform that feels truly premium and ready for production. The seamless integration of Clerk for authentication and Stripe for our tiered environment gives the app a very professional feel. On the backend, successfully building a custom architecture that normalizes chaotic financial data into a single cohesive feed is a massive technical win for our team. Accomplishing this level of high performance formatting at scale over a short time period is exactly what we set out to achieve.

What we learned

We learned the deep intricacies of interacting with financial data systems. We had to become familiar with CIK identifiers, SEC accession numbers, and XBRL parsing. More importantly, we learned how to engineer software effectively in the era of Generative AI. We learned how to act as orchestrators and architects directing AI agents like Claude and Antigravity to handle the heavy lifting of code generation. This hackathon dramatically shifted our perspective on what a small focused team can accomplish in record time when using the right AI tools.

What is next for Erns

Our next major step is integrating live LLM powered summarization of dense earnings call transcripts directly into the terminal, which will help explain market movements in plain English. We also want to expand our database to cover global markets beyond just the United States. Finally, we plan to officially launch the platform under our new brand and domain name ernsdata.com so we can begin onboarding waitlisted retail investors.

Built With

Share this project:

Updates