Voice AI Agents with Twilio - VAIT

We took on the Labourup AI Resume Builder challenge by experimenting with a multi-agent system designed to engage users after they submit a resume on our website. The goal was for the agent to place a call, conduct a natural conversation, and capture all the details needed to build a complete, structured resume that could be stored in our database. While our integration of the master OpenAI-based LLM agent with Twilio is still being fine-tuned, we’ve been able to move forward in the meantime by leveraging Twilio’s default conversation handler as a temporary solution. Our system is designed around a modular architecture with three core agents: a base agent, a conversation agent, and a resume-building agent. To streamline operations, we’ve developed APIs that coordinate the workflow between these agents. On the front end, we provide a web interface that allows users to easily submit their partially completed resumes, which then serve as input for the system.

Alternatives to calling agent providers that could be OpenAI’s Realtime API. Potential improvements include decreasing latency, adding a feature to enable users to call back the number in case they can’t pick up the call at the time, live updates on their profile as they speak to the agent, and automatic resume integration!