I’d like to propose a broader approach to OpenAI hardware: instead of building another standalone AI gadget or a closed family of proprietary devices, create a personal AI hub that acts as the center of an open ecosystem of devices from different manufacturers.
The key idea is simple:
ChatGPT is the brain. The hub provides continuity and connectivity. Everything else can become a peripheral.
The Personal AI Hub
The core device could be a small, screenless or minimal-screen personal hub with its own battery.
It could connect to the internet in several ways:
- Wi-Fi;
- a smartphone hotspot or tethering;
- optionally, its own eSIM/5G connection.
A separate cellular subscription should not be required. Some users may want a completely independent device with its own mobile connection, while others may prefer to use home Wi-Fi or share their phone’s connection.
The hub could support Bluetooth, Wi-Fi, UWB/NFC, and other suitable standards for communication with nearby devices, with some functions processed locally when appropriate.
But the hardware specifications are not really the main point.
The important part is the architecture.
The Hub Is the Brain — Other Devices Become Its Senses and Interfaces
Instead of requiring users to buy an entire OpenAI hardware ecosystem, ChatGPT could use compatible devices that already exist around the user.
For example:
Bluetooth headphones
Microphones become an audio input for ChatGPT, while the speakers provide its voice.
Smart glasses
Cameras provide visual context, microphones provide audio, and a display or lens projection system provides a visual interface.
Smartphones
They can provide internet connectivity, a display, cameras, sensors, additional computing resources, or simply act as another interface.
Computers
They provide a full-size workspace, applications, files, and potentially agentic/computer-use capabilities.
Cars and smart-home devices
They become additional places where the same personal AI is available.
The user chooses the hardware.
ChatGPT remains the intelligence layer connecting it all.
One AI, One Continuous Context, Many Devices
To me, this is potentially more important than the physical design of the hub itself.
Imagine starting a conversation with ChatGPT through your headphones while walking.
You get into your car and continue the same conversation through the vehicle.
You arrive home and continue through another device.
Then you sit down at your computer and the same AI session becomes a full workspace with a monitor, applications, and files.
The input/output device changes.
The AI, identity, and context do not.
Instead of opening ChatGPT separately on five devices, the user effectively carries one persistent personal AI that can move between available interfaces.
Smart Glasses Could Become the Visual Layer
AR glasses would make this architecture particularly powerful.
Imagine looking at a piece of equipment and asking:
“Which component should I check?”
ChatGPT could answer through the headphones while also highlighting the relevant component directly in the user’s field of view.
The same principle could be used for:
- equipment repair and maintenance;
- professional and technical work;
- real-time translation;
- navigation;
- identifying objects;
- contextual instructions;
- accessing project information while working;
- everyday visual questions.
The user would no longer need to repeatedly take out a smartphone, open an app, take a picture, upload it, and explain what they are looking at.
The AI could already share the user’s immediate context — subject to explicit permissions and privacy controls.
This Does Not Have to Replace the Smartphone
I don’t think the goal necessarily needs to be “replace the smartphone.”
A smartphone currently combines many different functions into one object: modem, computer, camera, display, microphone, sensors, identity, and applications.
A personal AI architecture could separate those functions.
The hub provides identity, connectivity, continuity, and access to the AI.
Headphones provide hearing and voice.
Glasses provide vision and visual output.
Computers provide large displays, applications, and productivity tools.
A smartphone can remain part of the system and provide whatever resources are useful — including its internet connection.
Over time, users may simply become less dependent on the phone as the central interface.
Why I Think an Open Ecosystem Matters
This is probably the most important part of the concept.
I would strongly prefer this not to become another ecosystem where the full experience requires:
OpenAI headphones + OpenAI glasses + OpenAI watch + OpenAI hub.
OpenAI could instead define a secure device layer, protocol, or certification standard that allows third-party hardware manufacturers to expose specific capabilities to ChatGPT.
A device might say, in effect:
I have two cameras.
I have three microphones.
I can play audio.
I have a small visual display.
Here are the capabilities the user has allowed ChatGPT to access.
Permissions could be granular and transparent.
This could allow manufacturers to build ChatGPT-compatible headphones, glasses, vehicles, cameras, home devices, and even specialized professional equipment.
In that model, OpenAI would not need to manufacture every possible device.
It could provide the intelligence layer and the standard connecting them.
The Bigger Idea
I realize that individual parts of this concept may already exist, have been proposed by others, or even be under development internally.
I’m intentionally not trying to specify the exact processor, battery, radio configuration, form factor, or other engineering details. OpenAI’s hardware engineers are obviously better positioned to make those decisions.
The idea I wanted to contribute is the architecture as a whole:
**Personal AI Hub
-
flexible connectivity (Wi-Fi / phone / optional cellular)
-
continuous AI identity and context
-
open third-party peripherals
-
headphones as audio interface
-
AR glasses as visual interface
-
phones, computers, cars, and homes as additional resources
= one persistent personal AI across the user’s environment.**
Instead of asking:
“What should the OpenAI device be?”
perhaps another useful question is:
“What if the OpenAI device became the hub that allows many devices to become part of ChatGPT?”
If this concept is useful to OpenAI’s product or hardware teams, I’d be happy to discuss it further or contribute a user/product perspective.
And if, a few years from now, I simply receive a message saying, “Thanks, Ivan — we built it,” I’d consider that a pretty great outcome too. 
— Ivan