OpenAI president Greg Brockman has offered a rare glimpse into what comes after the chatbot era—at least from the company’s perspective. In a recent interview with Joanna Stern on her YouTube channel, Brockman described OpenAI as “building a family of devices” designed to help people interact with its AI models. The phrasing matters. It suggests something broader than a single gadget or a one-off experiment. Instead, it points toward a platform strategy: multiple form factors, multiple entry points, and a consistent way for users to ask questions, get help, and carry out tasks using OpenAI’s systems.
What Brockman did not do is equally revealing. He didn’t confirm specific rumors that have been circulating for months—most notably the idea that one of these devices could be a smart speaker launching as early as 2027. He also didn’t validate earlier speculation that the company might pursue a wearable concept similar in spirit to Humane’s AI pin. And he didn’t provide a timeline beyond a general promise that people should expect these devices “soon.”
That combination—direction without confirmation—creates a familiar tension in tech reporting: the public wants details, but companies often keep their most consequential decisions behind closed doors until they’re ready to ship. Still, Brockman’s comments are substantial enough to map out what OpenAI likely believes about the next phase of AI adoption, why hardware is suddenly central to that belief, and how Apple’s ongoing legal fight with OpenAI could intersect with the company’s device ambitions.
A “family of devices” is not just a product plan—it’s a user behavior plan
When a company says it’s building a “family of devices,” it’s usually shorthand for a deeper thesis: that AI won’t become truly useful at scale through one interface alone. Chatbots are powerful, but they’re also context-limited. They live in apps, websites, and screens—places where users must remember to open them, type into them, and then interpret responses in a way that fits the moment.
Devices change that equation. A device can be always-available, always-listening (in some form), and always embedded in the physical routines of daily life. That means the AI doesn’t just answer questions; it becomes a companion to actions. It can help with planning before you leave the house, assist while you’re cooking, translate or summarize while you’re reading something in the real world, or guide you through tasks without requiring you to stop and “start a chat.”
Brockman’s wording implies OpenAI is thinking in terms of distribution and friction reduction. If the goal is to make AI feel less like a tool you summon and more like a capability you rely on, then hardware is one of the most direct ways to reduce friction. A family of devices also allows OpenAI to cover different user preferences and different environments. Some people want a voice-first experience. Others want a screen. Some want something wearable. Others want something that sits in the home and becomes part of the background.
Even without confirming any particular rumor, the “family” framing suggests OpenAI is preparing for a world where AI interactions happen across multiple surfaces rather than being confined to a single app.
Why the smart speaker rumor keeps coming back
Smart speakers have been a recurring target for AI integration for years, and the reason is straightforward: they’re already built around voice interaction. They’re also already positioned in the home, where many everyday requests occur—setting reminders, answering questions, controlling other devices, and providing quick assistance.
The Verge previously reported on rumors that OpenAI’s rumored smart speaker could arrive as early as 2027. Brockman didn’t confirm that. But his refusal to deny it outright doesn’t mean the rumor is accurate either. What it does suggest is that the company is at least considering the kinds of devices that naturally fit voice-based AI.
There’s also a strategic angle. A smart speaker is not only an interface; it’s a data and feedback loop. It can capture patterns in how people ask for help, what they ask repeatedly, and what kinds of responses are most useful in real time. That information can improve models and interaction design. It can also help OpenAI understand which capabilities matter most to users when they’re not actively “testing” the system.
However, smart speakers come with challenges too—privacy expectations, wake-word reliability, and the need to handle ambient noise and household variability. If OpenAI is serious about a device family, it likely has to solve these issues in a way that feels trustworthy to consumers. That’s not impossible, but it’s not trivial.
The wearable speculation—and why it’s plausible even if unconfirmed
Earlier rumors suggested OpenAI might pursue a wearable concept, potentially in the direction of a device like Humane’s AI pin. Again, Brockman didn’t confirm this. But wearables remain a logical extension of the same underlying thesis: AI should be available without requiring you to pull out a phone or open an app.
Wearables also offer a different kind of interaction model. Instead of asking users to initiate a conversation, a wearable can interpret signals—what you’re looking at, what you’re doing, what you might need next. Even if the device doesn’t fully “understand” everything in the environment, it can still reduce the gap between intention and action.
The challenge is that wearables are harder to get right than speakers. They require comfort, battery life, reliable connectivity, and a user experience that doesn’t feel distracting. They also raise higher stakes for privacy and consent because they can be perceived as more intrusive. If OpenAI is building a family of devices, it may be exploring multiple categories precisely because each category solves different problems and introduces different risks.
Brockman’s “soon” comment suggests OpenAI is moving beyond purely experimental prototypes. But “soon” in consumer hardware can still mean a long runway of engineering, manufacturing readiness, and software integration. Wearables, in particular, often take longer to mature because the physical constraints are unforgiving.
The real story: OpenAI is trying to reinvent the computer interface
Brockman’s interview wasn’t framed as a simple announcement. It was framed as a reinvention of how people interact with computers. That’s a subtle but important shift. For years, the dominant AI interface has been the chat window. It’s effective, but it’s also a continuation of the same interaction paradigm that existed before AI: type, submit, read.
If OpenAI is building devices, it’s likely because it believes the chat window is not the final form of AI interaction. Devices can support richer input and output modalities—voice, gestures, cameras, sensors, and ambient displays. They can also enable continuous assistance rather than discrete Q&A.
This is where the “family” concept becomes more than marketing. A family of devices can represent different interaction modes that collectively approximate a new interface standard. One device might be optimized for voice and quick tasks. Another might be optimized for visual understanding. Another might be optimized for portability. Together, they create a more seamless experience across contexts.
In other words, OpenAI isn’t just building products. It’s building a new way to use AI.
How Apple’s lawsuit could matter—without necessarily stopping hardware
Brockman was also asked about whether Apple’s lawsuit against OpenAI could impact the company’s device plans, particularly given OpenAI’s work involving former Apple designer Jony Ive. Brockman’s response emphasized focus on current work, though he didn’t provide details about timing or how legal developments might affect hardware.
This is a key point for readers: legal disputes don’t always halt product development. Sometimes they slow certain aspects of engineering, constrain hiring, or force changes in documentation and workflows. Other times, they primarily affect future negotiations and licensing rather than immediate shipping timelines.
Still, hardware projects are complex and involve many dependencies—industrial design, supply chain relationships, component sourcing, firmware development, and user experience design. If a lawsuit creates uncertainty around intellectual property, trade secrets, or specific technical approaches, it can introduce risk into the engineering process. Even if OpenAI continues building, it may need to adjust certain components or interaction designs to avoid exposure.
Brockman’s emphasis on staying focused suggests OpenAI believes it can manage those risks while continuing to develop devices. But it also implies that the company is not going to publicly speculate about legal outcomes or how they might reshape the roadmap.
For consumers, the practical takeaway is that hardware timelines can be influenced by legal uncertainty even when companies keep working. The absence of a confirmed release date from Brockman may reflect both engineering realities and the need to avoid committing to specifics under uncertain conditions.
What “soon” likely means in the device world
It’s tempting to interpret “soon” as a near-term launch. But hardware timelines are rarely that simple. Even if prototypes exist, turning them into mass-producible devices requires validation across manufacturing, reliability testing, regulatory compliance, and software polish.
There’s also the question of ecosystem readiness. A device family needs more than hardware. It needs a coherent software layer: account systems, model routing, safety policies, latency management, offline behavior, and integration with other services. It also needs a user experience that feels consistent across devices. If OpenAI is building a family, it likely wants the experience to feel like one system rather than separate gadgets.
So “soon” could mean that announcements, demos, or limited releases are approaching. It could also mean that the company is nearing a stage where it can share more concrete details without undermining the product’s competitive position.
The unique angle here is that OpenAI appears to be treating devices as part of the AI product itself, not as a side project. That typically requires a level of coordination that takes time.
Why this matters now: AI is moving from novelty to infrastructure
The broader context is that AI is shifting from novelty to infrastructure. Early AI adoption was driven by curiosity and experimentation. People tried chatbots because they were impressive. But the next phase depends on whether AI becomes dependable and integrated into everyday workflows.
Devices are one of the fastest ways to make AI feel dependable. They can provide low-friction