Aurbit

As a passionate distance runner, I've noticed that smartwatches like my Garmin watch track tons of health metrics. However, I've also found the vast amount of data to be overwhelming, and I know that most athletes don't know how to interpret this data or use it to improve. I found the perfect opportunity to contribute to a solution with Aurbit, a startup founded by two professional runners for Nike who noticed the same problem. The app features specialized AI agents that work together to guide your training.

As an agentic AI developer for Aurbit, my current focus is working on an injury-prevention AI agent that analyzes health metrics, including heart rate variability, running form data, and sleep quality to identify injury risk patterns before they become problems. The agent synthesizes data using OpenAI's function calling with a function calling architecture, where specialized tools retrieve and analyze data on demand. The agent decides which tool to call based on tool descriptions and the context of the prompt, synthesizing results to provide personalized recommendations while keeping token usage efficient.

Aurbit currently offers three agents with different specializations. I'm working on adding an injury prevention agent.