AI & technology 2:20 · Transcript included
Nikita Le Messurier, GTM Digital Natives at OpenAI, on why anyone who can speak is now a coder.
Nikita doesn’t write code in any traditional programming language, yet she builds every day by talking to AI. We talk about why voice is her favourite way to work right now, how real-time voice models make talking to AI feel like a natural conversation, and why the founders of this era have never had it so good.
Interviewed by Jess Wilson · What’s Next in Tech series · The Pillars, Sydney · September 2026
Three things that stood out.
- 01If you can speak, you can code. Anyone who can put words into Codex, spoken or typed and in most languages, can now build software. Nikita can’t code in the languages that came before, but she can “code in English”.
- 02Voice is the new interface. Real-time voice models don’t stitch speech-to-text and text-to-speech together. They run in parallel, so you can talk over the AI, change your mind mid-sentence and get things done much faster, the way two people talk.
- 03Building has never been faster. I built an app ten years ago. Today, getting from idea to MVP to scale is within reach of anyone with a voice, and Nikita is literally coding while she walks.
Full transcript
Lightly edited for readability.
Nikita: If you are able to speak or use your voice, or you can type, and somehow put words into Codex, you are now a coder. In most languages as well, which is really exciting.
Jess Wilson: When it comes to the different workflows and how we achieve work in this day and age, there’s also different modalities. Can you speak to us a little bit around the different modalities of work that OpenAI is releasing and working on?
Nikita: So my favourite at the moment is voice.
Jess Wilson: Yes, I’m absolutely obsessed.
Nikita: Yes, so obsessed with voice. So I don’t code. I can’t code in any particular coding languages that have been historic, but I can code in English, which is wonderful.
Jess Wilson: Isn’t that wild though? Because I built an app ten years ago, and that compared to now, I’m like, the entrepreneurs of this era don’t know how well they have it.
Nikita: No. When it comes to building an MVP and then, you know, scaling something out, I mean if you are able to speak or use your voice, or you can type, and somehow put words into Codex, you are now a coder. In most languages as well, which is really exciting.
Nikita: So a couple of weeks ago we released ChatGPT Live, our real-time voice model, which is really exciting. That’s where you’re able to actually interface via voice in real time with ChatGPT. Again, you can also embed this into your product for your customers as well, using the real-time voice models.
Nikita: But essentially, instead of having speech-to-text and text-to-speech kind of stitched together, they’re both operating in parallel. And so you’re able to have this really natural conversation with the agent, and it works perfectly. Like, I’ll be chatting to it and I’ll say, “Oh, by the way, can you create me X, Y and Z and then spin up Y,” and as I’m talking it’ll just talk over me and say, “Oh Nikita, but you know, did you mean X or…” You know, in the same way that you and I might have a conversation, where there’s a little bit of repartee sometimes, a bit of crossover.
Nikita: And to build that capability in without having the long pauses for the thinking, and the translation from speech-to-text or text-to-speech, is amazing. The speed at which you’re able to then have a conversation and get things done is so much faster. And so now I’m coding while I’m walking, which is absolutely wonderful.