We wanted to answer: "What if we could simulate thousands of financial scenarios AND have AI agents debate the results?" Traditional models give single-point estimates—we wanted the full picture with built-in adversarial validation.
What We Learned Orchestrating multi-agent AI debates (Gemini vs DeepSeek) that converge toward consensus Mathematical grounding to prevent AI hallucinations in financial contexts Running Monte Carlo simulations (10,000 scenarios) for robust uncertainty modeling
How We Built It Landing AI ADE → Extract financial data from PDFs Monte Carlo Engine → Simulate 10,000 scenarios with adjustable parameters AI Debate System → Gemini (optimist) vs DeepSeek (critic) with convergence detection Interactive Dashboard → Real-time analysis + PDF report generation
Log in or sign up for Devpost to join the conversation.