APEX | Devpost

Inspiration

Most people want to invest, but feel overwhelmed, uneducated, or unsure who to trust. Robo-advisors are black boxes and traditional financial education is confusing and slow. We wanted to create something that teaches users while they invest, instead of forcing them to choose between automation and understanding.

What it does

APEX is a desktop AI financial operating system that uses a human-in-the-loop multi-agent framework. Users can watch AI agents debate market conditions, strategies, risk, and execution, and actively participate via voice or text before any trades or recommendations are made.

How we built it

Multi-agent architecture (Market, Strategy, Risk, Executor, Explainer + User Agent)
Real-time financial feeds (news sentiment, volatility tracking, macro indicators)
Alpaca trading API integration for secure execution
Local-first desktop UI focused on transparency and explorable reasoning
Conversation layer that adapts explanations to user knowledge level

Challenges we ran into

Designing a system where agents communicate in a way that is transparent and interpretable without overwhelming the user
Balancing autonomy vs. user control (human-in-the-loop logic)
Ensuring risk controls activate even when strategies are aggressive or user overrides occur
Creating explanations that feel natural, not like generic financial textbooks

Accomplishments we’re proud of

Built a full agent discussion interface where reasoning is visible in real-time
Achieved plain-language explanations of complex financial concepts
Created a workflow where beginners gain understanding instead of blindly following automation
Made investing feel collaborative, not isolating or intimidating

What we learned

Transparency is not just a “nice-to-have,” it’s the difference between trust and confusion
Users learn best when they can interrupt, question, and override
Financial AI should be built around guidance, not authority