AI is everywhere. Whether we’re talking about Apple Intelligence, Genmoji, Siri, Gemini, or ChatGPT, it’s hard to avoid the sudden infusion of bots on our iPhones, Macs, and just about every other device we use.
And it’s not about to stop anytime soon. Apple Intelligence is set to have a huge year in 2026, with the new and improved Siri finally getting its moment in the spotlight. In fact, Apple is rumored to be partnering with Google as the foundation for the new Siri, with a custom large language model with a reported 1.2 trillion parameters. That’s a lot of Gemini.
With that in mind, I recently turned to Google’s chatbot to see what it knows about the iPhone, specifically the upcoming foldable iPhone. While the results were impressive and fantastical, it’s a far cry from anything Apple will release next year. But it gives a good glimpse into what Gemini can do and how it thinks. Let’s dive in.
The Gemini iPhone: Features and specifications
Before I cover the prompts I used to design a foldable iPhone, it’s important to point out that the process is usually much more elaborate. Designers hold focus groups, make “mood boards” to capture colors, text, and formatting ideas, and then develop detailed design specs. It’s not an easy process and can often take weeks or even months. I remember visiting Frog Design — the design lab famous for working with Apple in the early days — and talking about the highly technical process they use to determine user intent.
I mention all of that because Google Gemini seems blissfully unaware of what it actually takes to design a product. The bot was happy to take my basic design concepts and dive right into creating an AI-generated mock-up. That said, Gemini is helpful for quick images to help you see what the final product might look like.
I started by listing a few basic specs and asked Gemini to include a holographic display that emits from a tiny projector.
Here are the prompts I used:
Design a foldable iPhone
Make it thinner than the iPhone Air
Use two displays that fold — about eight inches diagonally
Project a holographic display onto a table
Using a tiny projector lens
Of course, you can elaborate even further and write an entire page of instructions. Next, I asked Gemini to take my feature list and flesh it out more. I was impressed with the level of detail and excited to see my first design mock-up. Here’s part of the specs:
| Feature | Specification | Rationale |
| Holographic Projector | Micro-LED or DLP Projector with a tiny, integrated lens. | The core innovation, allowing for shared, interactive, and spatial computing experiences. |
| Projected Display | Projects a holographic image onto a flat surface (e.g., a table). | Allows users to interact with a 3D or 2D interface (e.g., a full-size virtual keyboard, shared game board, or large-scale document) beyond the physical screen size. |
| Camera (Internal) | Under-Display Camera (UDC), likely 24MP resolution. | Eliminates the notch/hole-punch for a truly seamless main screen experience. |
| Biometric Security | Touch ID integrated into the power button. | A reliable and space-efficient solution compared to implementing Face ID on a foldable screen. |
The Gemini iPhone: First design mock-ups
Gemini had listed the design specs and even named the device the Helios, which has a nice ring to it, though I prefer iPhone Fold myself. I then asked Gemini to show me the initial design mock-up. The first design iteration was not quite right, though — which is par for the course with modern chatbots. The design was far too thick, the projection didn’t look remotely possible, and in many ways, the device didn’t even look like an iPhone at all. And it didn’t actually fold.
Google Gemini made some educated guesses about the next iPhone at first.
Google Gemini/Foundry
Yet, I only had to type in a few more prompts to adjust the mock-up. For example, I asked Gemini to make the phone thinner than the current iPhone Air, and that worked wonders. I also had to tell Gemini to make the projector lens much smaller. Chatbots respond to specific commands. “Make the design thinner” doesn’t work as well as “make the design thinner than the iPhone Air” because there’s a reference point. The next design looked better:
Google Gemini made a valiant second attempt at this design, showing a foldable phone.
Google Gemini/Foundry
I wasn’t quite ready to accept the mock-up, and this is where you can really benefit from the instant gratification aspects I mentioned. I asked Gemini to start over with a new design, taking all of my previous comments into consideration. Lo and behold, Gemini created a mock-up that was quite impressive. It “felt” much more like an Apple product as well. I finally could live with the basic design for the holographic display, even if it wasn’t perfect.
The iPhone Fold design by Google Gemini, a third attempt.
Google Gemini/Foundry
The Gemini iPhone: Tweaking the design and the mock-ups
Once you have a design you like, it’s possible to then make minor adjustments — as long as you are specific about the prompts. For example, I asked Gemini to make the design “just a hair thinner” and that worked because Gemini understood my intent.
It was also easy to change the color. I used this prompt: “Don’t change anything about the design except the color of the phone. Make it all white not black or silver.” Amazingly, that worked instantly and I felt like the design was starting to take shape. I decided to have some fun with my design and asked Gemini to put the phone in a coffeeshop instead so I could envision what it would look like in a familiar setting. I asked Gemini to put people in the background as well.
Here’s what that mock-up looked like:
Google Gemini can change the color of a mock-up very quickly.
Google Gemini/Foundry
Gemini surprised me with my next request. I asked the bot to add Tim Cook, the CEO of Apple, to the photo. I was shocked because I didn’t think a Google product would take a recognizable person (even though he is a public figure) and add him to an iPhone mock-up image. However, Gemini was happy to oblige even if the result looked a bit comical. The phone was way too big, and the person in the photo didn’t really look completely like Tim Cook (who wears glasses).
Google Gemini mock-up of someone who kind of looks like Tim Cook holding an iPhone Fold.
Google Gemini/Foundry
What’s wrong with this process
Designing the next iPhone took me about 30 minutes, including all of the back-and-forth with Gemini asking for design tweaks. Anything worth doing and worth doing well takes time, though. Think of building a business. You can’t come up with an idea and launch a company in five minutes. You have to study your demographic audience, understand their motivations and desires, tap into their felt needs.
Creative work is the same. While I loved seeing instant results, I didn’t feel the final design relayed a working prototype of a legitimate product that could exist in the real world. Icons were too small, speaker holes were everywhere, the selfie camera is impractical, and the whole thing is just way too big. And none of the images showed a phone with an actual folding screen.
Yet, it all seemed like an illusion. What was missing was the full design process — what are the user expectations, how does it compare to other devices, and who will actually use it? My main takeaway is that Gemini is a fine way to envision initial sketches, almost like the napkin phase of design, where you are roughing out an idea. When I asked Gemini to list all of the design specs and features first, eventually, things started to take shape a bit more.
It’s hard to say how this will translate to a Gemini LLM on the iPhone, but it’s a look into why Apple has taken such a long time to deliver its next-gen Siri. AI doesn’t live in the real world — it is a figment of our imagination. And the Gemini-powered Siri will need to have a much firmer grasp of reality.
