Greg Brockman just dropped something important in an interview with Joanna Stern, and I don’t think people are reading between the lines enough. OpenAI isn’t just building AI models anymore. They’re building devices. A family of them, actually. And the president’s messaging is crystal clear: the keyboard is becoming obsolete.
When Brockman says we’ll shift to talking to our computers for the “vast majority of what we do,” he’s not speculating about some distant future. He’s describing OpenAI’s product roadmap. This isn’t hype. This is conviction from someone who sits in the strategy meetings.
The Keyboard Was Always a Placeholder
Think about it. The QWERTY keyboard exists because of typewriter mechanics. We’ve dragged this 150-year-old interface into the digital age because we didn’t have anything better. Voice was always the endgame for human-computer interaction. It’s more natural. It’s faster for most people. It’s more accessible. The only reason we haven’t switched is that the technology wasn’t good enough.
Until now.
What changed is large language models got smart enough to understand context, handle ambiguity, and produce reliable outputs. They can actually parse what humans mean, not just what we literally say. That’s the missing piece that makes voice-first computing viable at scale.
I’m watching this transition happen in real time across https://mgks.dev/tags/ai-interfaces/, and the implications are enormous. We’re not talking about Alexa telling you the weather. We’re talking about replacing the fundamental input method for how we interact with computers.
What This Means for Developers
Here’s what keeps me up at night: most developers aren’t preparing for this shift. We’re still building for keyboard and mouse. We’re still designing interfaces around visual hierarchy and clickable elements. That paradigm is about to get disrupted hard.
If the majority of computing becomes voice-first, then every application needs to be reimagined for audio interfaces. Think about what that means for your typical SaaS product. The entire UX design language changes. Discoverability becomes audio-based. Error messages need to be conversational, not visual alerts. Multi-step workflows need to sound natural when spoken, not efficient when typed.
The technical challenges are substantial. You need to handle interruptions gracefully. You need natural error recovery. You need voice-specific authentication that doesn’t suck. These aren’t trivial problems.
Brockman mentions that OpenAI is working on “verifiable, auditable guarantees” around privacy in these voice interactions. That’s a big deal because voice is inherently more intimate than text. People will be more hesitant to use voice AI if they don’t trust that their queries aren’t being logged or analyzed. The technical implementation of cryptographic proofs that data stays private while computation happens is genuinely hard. This is exactly the kind of trust infrastructure that needs to exist before voice becomes mainstream.
The Office Problem Is Real
I appreciate that Brockman acknowledges the awkwardness everyone’s thinking about: what happens when everyone in an open office is talking to their computer at once? The mention of “soundproof beak masks” is funny until you realize he’s right about it signaling product-market fit. People are already creating hardware solutions to make voice-first interfaces work in shared spaces. That’s actual demand, not speculation.
But this also hints at a bigger opportunity for developers. If voice is going mainstream, then spatial audio, acoustic privacy, and ambient computing become core competencies. There’s an entire category of problems to solve around voice interfaces that nobody’s really tackled yet. Check out https://mgks.dev/tags/emerging-tech/ for more on where the actual innovation opportunities are.
The fact that Brockman says these devices are coming “soon” without giving a timeline is classic tech-speak, but it tells you this isn’t vaporware. OpenAI is shipping something. Apple’s lawsuit gets a shrug because, frankly, OpenAI doesn’t need anyone else’s permission to innovate on hardware and voice interaction.
The Broader Shift
What I’m most interested in is how this changes the competitive landscape. Apple built an empire on the GUI. Google dominated through search. Whoever wins voice-first computing wins the next computing cycle. OpenAI sees that clearly. They’re not building a device for the sake of it. They’re building the interface that will make their AI models irreplaceable in people’s daily lives.
For developers, this is both threat and opportunity. Your keyboard-and-mouse skills will eventually become legacy knowledge. But anyone who learns to think in voice-first design, who understands how to build applications that work through conversation, will be valuable for the next decade.
The question isn’t whether voice becomes dominant. It’s whether you’re ready when it does.