Bud AI

This is a submission for the Presage SmartSpectra SDK Challenge

What I Built

Bud AI is an intelligent, multi-agent educational platform designed to provide personalized, bio-aware academic mentorship. By integrating the Presage SmartSpectra SDK, Bud AI monitors a student's real-time physiological states (such as heart rate, cognitive focus, and stress indicators) via standard webcam feeds. When the system detects high stress or cognitive fatigue, its multi-agent system dynamically adapts—pivoting from a rigorous technical Teaching Assistant (TA) to a supportive, patient conversational buddy that breaks down complex concepts into manageable, bite-sized steps.

Demo

Live App Demo: bud-ai-platform.vercel.app

Video Walkthrough: Watch the 2-minute Demo on YouTube (Demonstrating real-time biometric tracking and dynamic multi-agent persona switching)

Code

Public GitHub Repository: github.com/jouzia/Bud-AI

Setup Instructions:

Clone the repository: git clone https://github.com/jouzia/Bud-AI.git

Install dependencies: npm install (frontend) and pip install -r requirements.txt (backend).

Configure your Presage SmartSpectra SDK API keys and LLM endpoint credentials in your .env file.

Run the application locally and grant camera access for bio-responsive tutoring.

How It Responds to Human Signals

Bud AI leverages the Presage SmartSpectra SDK to track subtle non-invasive physiological metrics during study sessions:

Cognitive Load & Heart Rate Variability (HRV): The SDK tracks real-time pulse and stress fluctuations as students tackle challenging problem sets or coding assignments.

Dynamic Multi-Agent Routing: When elevated stress or frustration is detected, Bud AI's intent router automatically shifts the active agent persona from a strict evaluator to a comforting, scaffolded mentor that offers encouraging hints rather than overwhelming data.

Real-Time UI Adaptation: The interface adjusts its layout, typography pacing, and mascot states to match the user's focus level, promoting healthy study habits and preventing burnout.

Prize Categories

Best Thoughtful Real-World Use: Bud AI addresses the critical real-world problem of student burnout, frustration, and cognitive overload in digital learning environments. By blending multi-agent AI tutoring with non-invasive physiological awareness via the SmartSpectra SDK, it creates an empathetic learning companion that cares about the student's mental well-being just as much as their academic progress.

Consent and Data Handling

Camera Consent: Bud AI explicitly prompts the user for camera permissions before initiating biosignal tracking, requiring an explicit opt-in toggle from the student or parent.

Data Privacy & Telemetry: All video processing and biometric calculations are performed 100% locally on the client device through the SmartSpectra SDK. No video feeds, facial images, or raw physiological metrics are ever stored, recorded, or transmitted to external servers.

Compliance Statement: Bud AI is strictly an educational tool and learning companion. It does not perform medical diagnoses, clinical evaluations, or therapeutic treatments, and prohibits any form of covert monitoring.

Agent Session

Agent Session Transcript: View Curated Session on Dev.to Agent Sessions (Highlighting the integration of the Presage SDK event hooks with multi-agent routing logic)