Submitted. Human AI Dyad is in for the All Things Agentic Hackathon, Collaborative Partner track. It's an agent that evaluates the working partnership between you and your AI tools — interviewing you across reliance, verification, judgment, and skill development, then building a personalised Dyad Rebalancing Plan. It came out of my post-doctoral research at Michigan State on AI-Induced Skill Ineptness, which we tested across five workforce datasets covering 1.6 million professional records; the findings held, and then the limitations section pointed out that no instrument exists to measure the construct. So I built one. It had to be a conversation rather than a questionnaire, because self-report is distorted by the very mechanism the theory describes — a fixed scale inherits that distortion, an interviewer that follows up on a hedge does not. Built as a single Google ADK agent on Gemini 3.5 Flash, running on Cloud Run with Firestore as persistent memory, so it remembers you across sessions and shows a trend line rather than a snapshot.
Log in or sign up for Devpost to join the conversation.