How we built a realtime system for responsive voice AI in six months
GPT-Live enables continuous voice interaction with AI, using a turnless speech model and low-latency architecture for faster, more natural conversations.
Get smart on it
A voice AI system has been redesigned to enable real-time conversation by having the voice model listen and speak simultaneously, rather than waiting for a separate system to detect when the user has finished speaking. Previously, turn-based voice systems used dedicated turn detectors that had to guess when to hand off conversation control, creating delays and awkward pauses. The new architecture separates media flow from other processing tasks, keeping audio streaming smoothly while delegating deeper reasoning and tool use to happen asynchronously without interrupting the conversation. This required rebuilding how audio is transported, how the model maintains conversation context, and how inference operates continuously rather than in discrete turn sequences.
Related news
Meet Shepherd: An Open-Source Python Substrate That Lets Meta-Agents Fork, Replay, and Revert Any Agent Run
Long-running agents accumulate state that no transcript captures. A coding agent at step 10 holds edited files, a running dev server, installed packages, and a warm prompt cache. When it misreads a traceback and rewrites a file that was already correct, neither available recovery path is cheap: patching forward grows the context and the token bill, and restarting from step one re-pays every model and tool call while reproducing nothing exactly, because runs are non-deterministic. Jumping back t
How to Disable Gemini in Gmail and Google Docs
New AI toolbars and prompts are showing up in Google Docs and Gmail. If you don’t want Gemini’s help in writing documents and emails, here’s how to turn that stuff off.
