Building Velox came with several challenges. Here are two key hurdles we tackled:

1. Developing Prediction Models & Data Sources

   • Crafting accurate paper models and sourcing reliable data for AI-driven predictions was a major challenge.
   • We addressed this by iterating on multiple statistical approaches, testing them against historical data, and refining them based on real-time market performance.

2. Learning & Implementing the Agent Framework

   • Since this was our first time working with an autonomous agents framework, there was a learning curve.
   • We overcame this by diving deep into documentation, experimenting with small agent interactions, and gradually scaling up to a fully functional agentic system.

Through persistence and continuous learning, we successfully built a robust AI-powered DeFi system!