Foundr

application architecture

🧠 Inspiration

Nearly 90% of startups fail, with 38% running out of cash and 82% struggling with cash flow mismanagement. Many founders lack the financial literacy needed to interpret term sheets, project runway, or plan growth. Foundr was born to provide 24/7 AI-powered financial support, so no founder has to crash due to avoidable money mistakes.

🚀 What it does

Foundr is your 24/7 AI CFO. It helps startup founders:

Analyze financial documents through a RAG-based chatbot
Get financial answers via Retell voice AI
Track runway, burn rate, MRR in real time
Simulate funding scenarios with Monte Carlo tools
Learn key finance concepts via a bite-sized learning hub
Join discussions in a forum for founder-to-founder support
Stay informed with startup news and pitch events pulled via APIs

🔧 How we built it

Frontend: React + TypeScript + TailwindCSS
Backend: Node.js + Express + AWS Lambda + API Gateway
Document Q&A: AWS S3 (storage), AWS Textract (OCR), Pinecone (vector DB), GPT‑4 (RAG)
Voice Assistant: Retell SDK, with serverless logic triggering AI analysis via Lambda
Forum: DynamoDB for posts, replies, and likes
Feeds: SerpAPI for events, NewsAPI for financial headlines
Simulations: Monte Carlo calculators running client-side with financial logic

⚠️ Challenges we ran into

Setting up seamless communication between S3, Lambda, and Retell for voice-based financial Q&A
Ensuring the Pinecone vector store updated in sync with uploaded documents
Designing a UI that balanced complexity with clarity for non-technical founders
Managing token costs and API limits on GPT-4 and News/Event scraping tools

🏆 Accomplishments that we're proud of

Built a fully functioning RAG-based document Q&A system in under 48 hours
Integrated voice AI using Retell for 24/7 conversational finance support
Enabled real-time metric tracking and simulations with live data inputs
Created a working forum and learning hub to build a community of financially literate founders

📚 What we learned

How to deploy a scalable, serverless pipeline using Lambda, API Gateway, and S3
How to build a production-grade RAG system with Pinecone + GPT-4
The power of combining voice + text UIs for founder accessibility
That even simple simulations can dramatically improve financial understanding for startup founders