{"type":"video","version":"1.0","html":"<iframe src=\"https://www.loom.com/embed/454cad125dde4a039c794d4f263784b2\" frameborder=\"0\" width=\"1006\" height=\"754\" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>","height":754,"width":1006,"provider_name":"Loom","provider_url":"https://www.loom.com","thumbnail_height":754,"thumbnail_width":1006,"thumbnail_url":"https://cdn.loom.com/sessions/thumbnails/454cad125dde4a039c794d4f263784b2-00001.gif","duration":90.463,"title":"PubMed Research Assistant","description":"In this video, I will demonstrate how Tygos, an API, connects hosted vector data to large language models. Specifically, I will show you how we connect the PubMed dataset of 33 million research papers to an app that validates research claims using GPT 3.5. We will start by examining a recent claim about the benefits of cold showers and using our app to validate it. I will guide you through the process of selecting a personality type and submitting the claim for validation. We will compare the results with and without data, highlighting the advantages of using specific research and citing sources. This demo showcases the power of Tygos in providing concrete evaluations and synthesizing relevant information without the need for managing vector data pipelines."}