Finlexity

Architecture
Sample Dashboard - 1
Sample Dashboard - 2

Inspiration💡

Every day, brilliant financial analysts waste hours doing the tedious, soul-crushing work of digging through SEC filings, copying numbers, and entering them into dashboards/slides—instead of what they were hired to do: think. At Finlexity, we asked: What if the grunt work vanished? What if analysts could skip straight to the analysis, the insight, the strategy, the deal-making? That’s why we built a tool that helps automate the initial company profiling process and speed up financial due diligence!

What it does🚀

Finlexity is a one-click AI analyst that automates the entire initial company profiling process. It scrapes SEC filings, extracts KPIs, margins, revenue streams, and more, then structures the data into polished dashboards and investor-ready reports. No more manual hunting, no more copy-pasting, no more wrestling with fragmented AI outputs. Just enter the target firm's name, click once, and get back to real analysis. Finally, finance professionals can focus on what matters: making smarter decisions, faster.

How we built it🛠️

We architected a multi-agent RAG system that would turn the financial intelligence game upside down: The Pipeline:

SONAR Deep Research APIs: 8 simultaneous calls gathering multi-dimensional company data
Agentic RAG Architecture: 5 specialized AI agents (Metric, Revenue, Countries, Goals, Team extractors) working collaboratively, leveraging SONAR APIs.
Vector Database: Semantic search powering contextual understanding
Dynamic Generation: PDF reports, JSON APIs, and interactive dashboards built over sleek immersive frontend.

Tech Stack ⚙️

Frontend: Next.js, React, Tailwind CSS
Backend: Python (Flask)
Document Parsing & Analysis: PyPDF2, LangChain, Perplexity Sonar API
Report Generation: ReportLab (for PDFs formatting)
Data Pipeline & RAG: Custom pipeline with retrieval-augmented generation (RAG)
Vector DB: ChromaDB
Deployment: Local environment for development

Challenges we ran into

Agent Coordination Chaos: Getting 5 AI agents to collaborate without stepping on each other was quite the challenge. We had to design a sophisticated orchestration layer.
Scale vs. Speed: Balancing comprehensive analysis (8 API calls + agent processing) with user expectations of instant results. We optimized our pipeline for the best possible latency despite Perplexity rate limits bottlenecks.
Financial Data Accuracy: Unlike chatbots, financial analysis has zero margin for error. We built multiple validation layers and cross-referencing systems.

Accomplishments that we're proud of 🏆

Cool Architecture: While everyone else builds basic document parsers, we created a multi-agent financial intelligence system. This isn't just AI - it's AI that thinks like a financial analyst team.
Production-Ready Output: Our system generates abridged FDD reports that jump starts analysts in the FDD process. The dynamic dashboard generated serves as a quick look up tool to analyze important metrics/info.
Scalable Innovation: Built to handle everything from startup analysis to Fortune 500 due diligence.
Speed Revolution: What takes analysts hours/days now happens in under 7 minutes with great accuracy and comprehensive coverage.