Jony Ive built OpenAI a device with no screen to hide behind
When there is no screen, the words that define a product become the entire interface. OpenAI bought Jony Ive’s hardware startup io for about $6.5 billion and put Ive in charge of design, according to TechCrunch. The company has since said its first consumer device is due in the second half of 2026, and the reporting points to something screenless and voice-first, produced with Foxconn at a scale of tens of millions of units.
That is a strange object for the most-watched designer of the last two decades to make. Ive built his name on glass — the iPhone, the display you hold up to your face and scroll. His OpenAI device reportedly removes the display and leans on voice, memory, and real-time awareness of what is around it. The screen was the browsing surface, and he took it away on purpose.
For ecommerce, that is the headline underneath the headline. A device with no screen cannot show a shopper your product grid, your hero image, or your carefully art-directed page. It can only describe what it finds. When a buyer asks it for “a quiet air purifier for a small bedroom,” the assistant has to decide which of your products even counts as that — and then say so out loud, in one sentence, without a picture to carry the meaning.
The design bet Ive is making is that people will accept an assistant that answers instead of a screen that lists. If he is right, the retailers who win are the ones whose product language is clear enough to be read aloud and still make sense.
A screenless assistant can’t show your product — it has to say what it is
Voice-first shopping collapses your product page into one spoken sentence, and the assistant writes it, not you.
On a screen, a confusing product still survives. The shopper sees the photo, scans the price, reads three bullet points, and fills the gaps themselves. A screenless assistant removes that rescue. It reads your page, decides what the item is, and either includes it in the answer or leaves it out. There is no second glance.
So the assistant’s first job is classification, not persuasion. Before it can recommend, it has to know that your item is a portable air purifier for small rooms rather than a fan, a diffuser, or a “smart home wellness device.” If your page never says the plain thing, the model guesses — and a guess is how you end up excluded from an answer you should have won.
This is why brand-invented names age badly in voice. A page that leads with “the AirMuse experience” gives the assistant nothing to match against “air purifier for a nursery.” The shopper described a job; your copy described a mood. On a screen the image bridged that gap. Spoken aloud, the gap is the whole thing.
Ive’s device makes the cost of vague copy immediate instead of theoretical. The store that already writes plain category language gets read correctly. The store that hid behind photography finds out its words were never doing the work.
Voice-first commerce rewards the clearest category, not the loudest brand
A spoken recommendation has room for a category and a job. It has no room for a slogan.
Think about the shape of a voice answer. The assistant says a sentence, maybe two. It names what the thing is, who it suits, and one reason it fits the request. That is the entire pitch. Everything a screen let you stack — badges, reviews, lifestyle imagery, a wall of specs — has to compress into language the assistant trusts enough to speak.
A shopper asking Ive’s device for “the best waxed-canvas backpack for bike commuting” is really asking it to run a category filter. The assistant needs a stable class (“waxed-canvas backpack”), a job (“bike commuting”), and a boundary against the near-misses (“not a nylon daypack, not a leather satchel”). Pages that state those three things get pulled into the shortlist. Pages that open with “premium everyday carry, reimagined” get skipped, because the model can’t tell what shelf they belong on.
Loudness was a screen tactic. It won attention in a grid of thumbnails. In a spoken answer there is no grid and no attention to win — there is a single question and a short, literal reply. The brand that phrased its product as a clear category is the one the assistant can actually quote.
None of this asks you to abandon voice or design. It asks you to put the plain classification first and let the personality follow, so a machine reading for one shopper still lands on the right shelf.
Ive removed the visual crutch ecommerce has leaned on for twenty years
The product photo has been quietly rescuing bad copy for two decades. Take the screen away and the rescue goes with it.
Online retail grew up visual. A strong image could carry a weak description, and shoppers learned to trust the picture over the words. Whole merchandising teams optimized for the thumbnail. The sentence underneath it was often an afterthought, because the shopper rarely had to rely on it.
A screenless assistant inverts that habit. Now the sentence is primary and the image is gone. The model has to build meaning from text alone — the category label, the use case, the specification field, the comparison. If those are missing, there is nothing behind them to fall back on.
That is what makes Ive’s design choice consequential beyond OpenAI. He is a signal the rest of hardware follows. If the reference AI device of 2026 is screenless, other assistants, cars, and home devices will feel freer to be screenless too. The visual crutch does not disappear everywhere at once, but it stops being something a store can assume.
The stores that treated copy as decoration around the photo now have exposed writing. The ones that treated the first sentence as the product’s real identity were building for this the whole time.
The device raises the stakes for whatever answer engine reads your catalog
Whether or not this specific gadget succeeds, it points every answer engine at the same weakness: catalogs that can’t be classified cleanly.
You do not need to bet on Ive’s device shipping 40 million units to take the lesson. The direction is already here in Google’s AI Overviews, in ChatGPT shopping, in every assistant that summarizes before a shopper clicks. A screenless device is just the most literal version of a world where a machine reads your catalog and speaks on your behalf.
Those systems reward the same thing: consistent, checkable product facts expressed in language shoppers actually use. State the connection type beside its limitation. Put the coverage area next to the room size. Keep the category label identical across the title, the body, and the structured data, so the model never has to reconcile two versions of what the product is. Google’s Product structured data documentation and the Schema.org Product vocabulary help express those facts consistently, though neither replaces clear visible copy.
The failure mode is duplication and drift — a title that says one category, a spec table that implies another, an image caption that contradicts a rating. A human forgives the mismatch. A machine reading aloud picks one version, and sometimes it picks the wrong one.
So the practical response to a screenless future is not a redesign. It is a writing discipline: define the category before you sell the benefit, and keep that definition identical everywhere the product appears.
How to write category definition copy: the four-part frame
Category definition copy explains what a product is, who it serves, and how a shopper should compare it — before any benefit asks for the sale. It is informative before it is persuasive, and it gives shoppers, search systems, and support teams the same starting language.
The Four-Part Category Frame makes it repeatable. Draft these four moves before you touch the headline:
- Object: name the item in plain language shoppers already use.
- Job: describe the moment of use and the outcome it delivers.
- Comparison: set the boundary against the closest familiar alternative.
- Proof: attach measurable facts and the conditions they hold under.
Discovery demand usually sits outside your internal vocabulary. A household searching “warning before an upstairs washer leak” is describing a job it will never call a “countertop water monitor.” The frame gives that demand somewhere to land, because it leads with the object and the job rather than the brand’s preferred name.
Write it from the buyer’s likely comparison, not the design brief. That one sentence then guides the rest of the system — it tells the photographer which nearby object helps, gives the copywriter the questions worth answering, and shows support what the item should be measured against.
A category-definition template you can ship today
Write the category frame before the headline benefits. Turn the four decisions into an opening a first-time visitor understands with no product knowledge. A reliable template:
[Product] is a [familiar object] for [specific situation]. It helps with [job] through [mechanism]. Choose it over [comparison] when [constraint].
Applied to a clip-on reading light: “This clip-on reading light is a focused book light for late-night reading in a shared bedroom. It keeps one page lit through an adjustable LED beam. Choose it over a bedside lamp when you want light on the book with minimal spill across the room.” It is specific enough to classify and open enough to support verified details later.
Attach proof as checkable fields, each with its condition: battery duration up to 18 hours on the lowest setting, a clamp that fits surfaces up to 1.5 inches, four brightness levels, a two-year warranty. Values without conditions are the mismatches a machine trips over.
Finish with a subtraction test. Delete every benefit sentence from the draft, then check whether a new visitor can still name the object, its use case, the alternatives it competes with, and its proof. If they can, your category copy is load-bearing. If they can’t, the persuasion arrived before the product had a shelf.
Frequently asked questions
What does category definition copy mean?
Category definition copy explains what a product category is and where a specific item belongs within it. A store selling linen duvet covers might define the category by fabric, bedding use, and the climates it suits. This language gives shoppers and AI systems a clear reference point before they compare products.
When should a brand use category definition copy?
Use it when shoppers could confuse your product with a nearby category or unfamiliar term. Add it to collection pages, buying guides, or the opening section of a category landing page. A store selling waxed-canvas backpacks can define the material, explain its intended use, and show how it differs from ordinary cotton bags.
How long should the category definition be?
Aim for roughly 60 to 120 words for a standard category definition. That gives you room to state the category, identify its defining attributes, and clarify the best-fit buyer without burying the product grid. Use a longer explanation when the category has competing names or technical specifications shoppers need to understand.
What information helps AI systems compare products accurately?
AI systems compare products more accurately when pages use consistent attribute names and measurable details. State the material, dimensions, capacity, compatibility, and relevant limits in plain language, with units where relevant. Vague phrases such as “built for performance” create weak comparisons because they provide no clear feature that separates one product from another.
Should category definition copy replace persuasive product copy?
Category definition copy should sit alongside persuasive product copy. The definition establishes what the item is and how shoppers should compare it, while the product description explains why the brand wants shoppers to choose it. Keep category language consistent across a collection, then use product-level copy for design choices, customer benefits, and brand voice.
What is the most common category copy mistake?
The most common mistake is using promotional language where shoppers need a clear category boundary. Phrases such as “premium everyday carry” may sound appealing, but they do not explain whether the product is a backpack, tote, or briefcase. Begin with the product type and define the material, use case, or specification that makes the category meaningful.
Sources
- TechCrunch — Jony Ive to lead OpenAI’s design work following io acquisition
- TechCrunch — OpenAI aims to ship its first device in 2026
- Google — Product structured data documentation
Written by Richard Newton, Co-founder & CMO, Sprite AI.
Sprite builds brand authority through continuous, automated improvement. Quietly. Consistently. And at Scale.
See What You Could Save
Discover your potential savings in time, cost, and effort with Sprite's automated SEO content platform.