TricherAI

Prototype concept: a secondary mobile device periodically (every ~10s) captures the laptop screen, performs image understanding & generates spoken answer guidance to the user over a discreet audio channel.

Pipeline sketch:

1. Capture → compress → secure upload.
2. Vision/OCR pass extracts question text & relevant options.
3. Retrieval + reasoning layer drafts an answer (with confidence scoring).
4. Low‑latency TTS streams audio response back to earphones.

Focus areas: latency reduction, hallucination mitigation & ethical guardrails.

Status: exploratory R&D – not for misuse; researching academic integrity compliant use cases (accessibility & revision assistance).