video1.0<iframe src="https://www.loom.com/embed/ccb88c40fb4f4e2b85cd9145a6612bce" frameborder="0" width="1778" height="1333" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>13331778Loomhttps://www.loom.com13331778https://cdn.loom.com/sessions/thumbnails/ccb88c40fb4f4e2b85cd9145a6612bce-75cfc0a8cdf5b9e6.gif401.36In this video, I walk through my implementation of my custom-built voice AI agent, detailing my approach to the problem and the technologies I used, including LiveKit, Deepgram, Llama.3.3 LLM, Twilio, and SendGrid. The voice AI agent successfully handles medical patient intake by asking a series of questions (including address validation), finds and schedules an user-selected doctor appointment, and sends an confirmation email. I faced challenges with tool calling and email validation, which I addressed through testing and documentation review. I also share insights on future improvements, particularly regarding email collection. Please let me know if you would like to demo the voice AI agent!