DeloadIQ

DeloadIQ is an AI strength-intelligence system that analyzes workouts, performance history, training videos, and recovery signals to decide when to deload, rebuild strength, and progress AI.

Comment

Inspiration

Strength training decisions are often driven by intuition, rigid programs, or generic deload rules. We wanted to build a system that treats training as a continuously evolving state, using real performance signals to decide when to push, hold, deload, or rebuild—rather than relying on fixed schedules or guesswork.

What it does

DeloadIQ is an AI strength-intelligence system that analyzes workout routines, session history, recovery signals, and training videos to maintain a persistent training state. It autonomously determines whether an athlete should push, hold, deload, or rebuild, and explains each decision with technical reasoning and actionable guidance. Video links are used for qualitative form and movement consistency assessment rather than real-time pose estimation.

How we built it

We built DeloadIQ as an interactive AI Studio app powered by Gemini 3. Workout data, recovery inputs, and video links are stored in-session to maintain longitudinal context. Gemini 3 Flash handles fast, frequent daily performance analysis, while Gemini 3 Pro with a thinking budget performs deeper, long-term training state reasoning and decision-making. The system re-evaluates its conclusions whenever new data is added.

Challenges we ran into

The main challenge was designing reasoning that works across weeks of training rather than single sessions. Balancing fast responsiveness with deeper longitudinal analysis required careful separation of lightweight and heavyweight reasoning paths.

Accomplishments that we're proud of

Built a state-based training intelligence system rather than a rule-based fitness app

Clear separation of fast vs deep reasoning using two Gemini models

Explainable outputs that show both technical reasoning and tactical recommendations

What we learned

Long-term performance analysis benefits from treating decisions as evolving states rather than static outputs. Separating frequent lightweight analysis from deeper reasoning significantly improves clarity, reliability, and scalability.

What’s next for DeloadIQ

Future work includes persistent storage across sessions, coach-level multi-athlete views, richer video-based movement analysis, and automated long-term progression planning across training cycles.

Built With

Share this project:

Updates