How We Built Our Multimodal Agent Harness
Adding images and live camera to a real-time voice loop is a systems problem. The engineering behind a unified multimodal pipeline: on-device frame resizing, one-frame-per-second throttling, and an auditable record of what the agent actually saw.
Is GPT-Live-1, OpenAI's Full-Duplex Model, Ready for Production?
The media path is simpler and the conversation is more natural. A technical breakdown of GPT-Live-1's architecture, its $0.05/min pricing, and the production controls that still belong outside the model.
Voice AI Observability In Production: What To Track On Every Call
What to track on every production voice call across system, audio, conversation, and operations layers, so failures get localized and routed to the right owner.
Voice AI Observability for Speech to Speech Models
Every voice AI deployment sits at a wall: "how do we know it's working?" This is how we built Lens to solve the observability gap in production speech-to-speech models.
What Full-Duplex Speech Models Actually Change in Voice AI
Full-duplex speech models eliminate the format conversion step where cascaded voice AI loses tone, emotion, and context. The architectural case, with real numbers.