InvestHER

Inspiration

It started with a conversation about a $10 cup of coffee. When early-career women discuss money, the number one stated blocker to investing isn’t a lack of financial literacy, it’s the genuine belief: "I don't have enough money to start."

The financial industry has a massive blind spot here. Women invest a significantly smaller share of their savings than men and start much later. Traditional fintech apps ignore this reality; they assume you already have thousands of dollars ready to invest, give you a lecture on compound interest, and leave you to figure out the budget yourself. We realized that to build true financial confidence, we had to solve the transaction friction first. We didn't want to build an app that asks users to find the money, we wanted to build an engine that finds the money for them.

What It Does

InvestHer is a mobile-first platform that turns everyday transaction data into your first investments. It connects a user’s daily spending habits directly to long-term wealth creation through three core pillars:

The Money Radar: A read-only engine that analyzes daily spending (like subscription overlaps or frequent coffee runs) to identify behavioral over-spikes. Instead of shaming the user with a red budget alert, it presents a frictionless "Found Money" card, allowing them to redirect that exact amount into an investment account with a single slider.
The LLM Financial Coach: An AI educator that explains complex financial concepts (like Roth IRAs or ETFs) using the user's actual found money numbers to show illustrative math, making the leap to investing feel personal and immediate.
Behavioral Circles: A gamified community leaderboard that rewards healthy financial habits—like learning streaks and budget check-ins—while keeping raw dollar balances and net worth structurally hidden to ensure total privacy.

How We Built It

We architected InvestHer to be scalable, frictionless, and completely decoupled.

Frontend: We built a warm, low-cognitive-load user interface using React and Next.js, styled to feel empowering rather than like a traditional, intimidating corporate bank portal.
Data Layer: To maintain a realistic hackathon scope while planning for enterprise scaling, we engineered a mock financial layer using a flat-file JSON structure. All data is isolated and routed behind a single interface function. This creates a clean architectural seam, meaning transitioning to a live production environment is a seamless plug-and-play with Plaid’s tokenized OAuth API.
AI Integration: We utilized the Claude API (via a Python backend proxy server) to power our LLM Financial Coach.

Challenges We Ran Into

Integrating Generative AI into a fintech product introduces massive regulatory compliance liabilities. If an AI crosses the line into providing unregistered financial advice, it triggers severe SEC Registered Investment Adviser (RIA) penalties.

We couldn't just wrap an API and hope for the best. To solve this, we treated the LLM's behavioral boundaries as a core security control. We adapted an LLM drift monitoring architecture to run systematic "poison pill" test cases against our backend proxy server. This ensured our system prompts actively prevented the model from regressing into unlicensed investment advice, gracefully pivoting users back to educational frameworks.

Furthermore, we had to navigate the SEC's strict guidelines on Digital Engagement Practices (DEPs). We spent a lot of time engineering our database schemas so that our gamification engine rewarded habits and structurally blocked the platform from gamifying trading volume or frequency.

Accomplishments That We're Proud Of

We are incredibly proud of building a zero-trust, read-only data architecture. By completely rejecting credential storage or screen-scraping methods, we ensured that even in a prototype phase, the design prioritized strict consumer data privacy.

We are also proud of the UX pivot. Changing the language of personal finance from "you broke your budget" to "we found $120 you could be investing" transforms a stressful banking audit into a highly rewarding, empowering discovery moment for our users.

What We Learned

We gained a deep understanding of the regulatory landscape surrounding fintech, specifically regarding SEC DEP guidelines and the technical limits of Reg S-P consumer data privacy. We also learned how to successfully decouple transaction parsing from our analysis engine, allowing us to build complex "found money" algorithms without getting bogged down by the messiness of raw bank data pipelines.

What's Next for InvestHer

Our immediate next step is to swap our architectural data seam with live Plaid and Fidelity OAuth endpoints to process real-time transaction enrichment.

Long-term, InvestHer is designed for clear B2B expansion. We plan to market the platform to enterprise employers who can purchase InvestHer as a corporate benefit, utilizing it as a powerful tool to attract, support, and retain early-career female talent while helping them build lifetime wealth.