{"type":"video","version":"1.0","html":"<iframe src=\"https://www.loom.com/embed/d24b08c8d1f448fc95192203f3230c3a\" frameborder=\"0\" width=\"1920\" height=\"1440\" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>","height":1440,"width":1920,"provider_name":"Loom","provider_url":"https://www.loom.com","thumbnail_height":1440,"thumbnail_width":1920,"thumbnail_url":"https://cdn.loom.com/sessions/thumbnails/d24b08c8d1f448fc95192203f3230c3a-5659d5a70153be84.gif","duration":888.912,"title":"StudyMind AI, Trustworthy RAG Study Assistant","description":"In my Loom, I demo Study Mind AI, a RAG-powered study assistant built to improve reliability and prevent hallucinations by grounding every answer in a student’s own notes. I built it with Python, GPT 4.1 mini, Chroma, and local sentence transformers, and added guardrails plus an automated evaluation harness. The pipeline is guardrails, then a StudyMind agent that classifies the task, retrieves top four chunks, generates an answer, and self-verifies with a confidence score and reasoning trace. I also ran 27 guardrail and test chunk cases, 11 predefined harness test cases, and showed prompt injection being blocked. No action was requested from viewers."}