Justin Uberti

any plans to extend the endpoints to other countries? domestic Indian SIP calls require the platform’s SIP and media termination to be in India.

Why is this not a checkmate?🤔😱

we know people love calling their dots - connection times are now much faster. also by popular demand - you can now hang up hands-free. just say “goodbye” to your dot and it’ll end the call.

On Day 1 of VapiCon, @juberti’s gives a talk on speech-to-speech models, “Skipping Turns: A More Responsive Architecture for Voice Agents.” Justin, Head of Realtime AI at @OpenAI is a hands-on engineer and vision-based leader of large (150+) engineering teams. He is also an occasional mathematician, physicist, and musician. Join Justin and other AI leaders in San Francisco this November. 📍 Fort Mason, San Francisco 📅 November 11–12, 2026 Explore the agenda and get your ticket: vapi.ai/vapicon-2026

What happens when the outbound agent reaches another voice agent? Do they stay in speech + DTMF, or could the call negotiate a structured exchange?

The interesting question is where vertical integration stops. Telephony and session management are natural extensions, but orchestration increasingly touches business logic, external systems, application-specific policies, etc. Do you see the model provider eventually owning that layer too?

New: OpenAI tells me 250 Million people talk with ChatGPT weekly (voice and dictation). About 20% of its user base.

So, does this mean OpenAI is going after the infrastructure layer(like LiveKit) and the orchestration layer(like pipecat). So theoretically you can have OpenAI and the Sip trunk layer(like Ozonetel,Twilio etc) and that's all you need for the voice agent stack. Interesting.

LLMs can mimic thought and feeling with uncanny skill, but intelligence alone is not enough for subjective experience go.nature.com/3VSXWQ4

My dear friends, I’d like to show you my latest toy project Introducing Handwired: turn your regular JavaScript code into a neural net! - no training - exact outputs vs the JS program - fully differentiable All runnable in a browser, without GPU

Real life, baby

you can now use native decision delegation in your voice app. a week ago i wrote this voice x decision doc. today i update the guide because now it is miles better you can trigger a decision on every input or output frame (roughly every 200ms). you can have a realtime decision engine judging what i’m saying while i’m still saying it!! try realtime action × realtime voice 👇 developers.openai.com/api/do…

New voices. Better accents. And less "checking..." GPT-Live-1.1 brings these updates to Voice in ChatGPT, Codex, and dots, with new voices in French, German, Hindi, Mexican Spanish, Korean, and Japanese. Rolling out today to all users. Coming soon to the API.

interpretability might be the most important scientific problem in the world right now, as I don't think we can solve alignment if we don't know what we're building. I think I have good news about it. I co-authored early work in mechanistic interpretability at OpenAI and came away pessimistic, because neural networks appeared to fold into arbitrary tangled knots. years later, I started openai's first large-scale automated interpretability project, finding about 2% of neurons legible with the primitive tools we had with modern agentic tools and a much closer look, the picture looks totally different. the vast majority in models up to trillions of parameters are legible, and their circuits beautiful. many famous examples of tangled neurons turned out to be misunderstood. The roadblocks I expected mostly aren't there. I now think ambitious interpretability will be solved by the time we need it most, and possibly quite soon, as automated scientists keep improving. it may be a bitter lesson for interpretability: the simple thing, at absurd scale, works already-clean.goodfire.com/

do you think that in the long term, a separate echo cancellation step (neural/linear) still makes sense, or just send raws(/+loopbacks) and let the model figure that out too?