Engineering voice agents: Latency, quality, and scale — Rishabh Bhargava, Together AI

Summary

This transcript focuses on engineering high-quality, low-latency voice agents at scale, with specific mention of Together AI's cloud infrastructure. Key points include the vast need for AI to handle existing human-executed phone calls and the potential of voice as a new interface for interacting with systems. The practical takeaway is that the pipeline architecture is becoming the dominant method for building production-ready voice agents, with ongoing evolution in the field.