Inspiration
Applying for an IPO often feels like gambling for everyday investors. Most market participants rely heavily on social media hype, brand recognition, and news cycles. However, the true financial health of high-profile offerings—such as hidden cash burn, related-party debts, and massive promoter exits through Offer For Sale (OFS)—is deeply buried inside 600-page legal prospectuses. Retail investors lack the time or training to audit these documents, leaving them vulnerable to market hype. ProspectusIQ was built to level the playing field.
What it does
ProspectusIQ is an institutional-grade AI forensic auditor that transforms dense regulatory IPO documents into clear, data-driven investment memorandums. Key features include:
- Real-Time IPO Screener: Automatically aggregates active and upcoming market listings without requiring manual entry.
- 12-Factor Forensic Rubric: Evaluates offerings across strict institutional dimensions like Capital Integrity, Earnings Quality, and Market Valuation.
- Divergence Detection Card: Contrasts public retail sentiment against fundamental risks extracted from actual filings.
- Institutional PDF Memorandum Export: Generates clean, publication-ready vector PDF reports with a single click.
How we built it
- Core Framework & UI: Built using Python and Streamlit to create a high-performance, dark-mode terminal experience.
- Document Processing: Utilizes PyMuPDF for targeted, selective chapter parsing of massive regulatory prospectuses.
- AI Engine: Powered by Google Gemini 1.5 Flash for rapid text synthesis, semantic analysis, and structured extraction.
- Data & Validation: Implements Pydantic for rigid schema enforcement and SQLite for deterministic local caching.
Challenges we ran into
- Handling massive 600+ page regulatory documents within context limits without losing track of critical financial data points.
- Designing a robust multi-tier fallback mechanism to guarantee uninterrupted extraction during high-latency states.
Accomplishments that we're proud of
Successfully building a multi-tier AI routing pipeline powered by Gemini 1.5 Flash that reduces dense 600-page legal prospectuses into structured 12-factor audits and professional vector PDF memorandums in seconds.
What we learned
We learned how to design deterministic financial audit pipelines using LLMs, structure selective text extraction to optimize token efficiency, and bridge the gap between complex legal jargon and intuitive retail visualization.
What's next for ProspectusIQ
Expanding automated data ingestion to cover global stock exchanges, adding historical post-listing performance tracking, and integrating real-time social sentiment stream analysis to widen the divergence detection engine.
Built With
- google-gemini
- pydantic
- pymupdf
- python
- streamlit