CrediFy

Inspiration

We've all seen it: the constant stream of "expert" financial advice on social media, often from unverified sources. From "to the moon" crypto predictions to guaranteed stock gains, retail investors are constantly bombarded with claims that can be misleading, biased, or outright false. This creates an environment rife with confirmation bias and echo chambers, leading many to make impulsive and often detrimental investment decisions. I build this tool which empowers everyday investors to cut through the noise, verify claims independently, and seek knowledge from trusted sources, fostering a more informed and resilient investing community.

What it does

The CrediFy is your personal AI-powered shield against financial misinformation. It allows users to:

Input any financial claim they encounter online, whether it's a tweet, a TikTok trend, or a forum post.
Optionally, provide a stock ticker for more focused analysis.
Receive an instant, AI-generated fact-check that includes:
- A clear verdict (e.g., "Verified," "Partially Verified," "Misleading," "Unverified").
- A concise summary of findings explaining why the claim holds up (or doesn't).
- Simulated citations to reputable sources, demonstrating how a real-world version would leverage Perplexity's trusted web-wide research.

This helps users quickly assess the credibility of financial advice and understand the underlying truths (or falsehoods) before making investment decisions.

How I built it

I built the CrediFy as a full-stack application, prioritizing a clean user experience and powerful AI integration.

Frontend: The Frontend was built using Next.js
Backend: The core logic resides in the backend, built with FastAPI and Pydantic for data validation. The backend architecture:

1.  It receives the financial claim and optional ticker from the frontend.

2.  Constructs and sends a targeted prompt to the **Perplexity Sonar API**.

3.  It parses the AI's response, extracting the verdict, summary, and citations.

4.  Finally, it sends this structured, fact-checked analysis back to the frontend for display.

Challenges I ran into

The main challenge was LLM reliability. Ensuring consistent, structured output from the Sonar API for accurate parsing proved a significant hurdle I tackled through rigorous prompt engineering.

Accomplishments that we're proud of

In a world where misleading news and outright scams can so easily lead to heartbreaking financial losses for everyday folks, it truly makes me happy to imagine people using this platform.

What's next for CrediFy

Enhanced Source Verification & Detail: We plan to refine the Perplexity API integration to provide even richer, more specific citations, potentially including direct links to SEC filings or company investor relations pages, boosting user trust.

Claim History & Watchlists: Future iterations will allow users to save past fact-checks and create watchlists for specific stocks or topics, enabling them to easily monitor for new, potentially misleading financial claims.

Community & Education Features: We envision adding features that allow users to report new trending claims for fact-checking and provide educational context on common financial terms, further empowering informed decision-making.