The new screens run on the web. Glasses, XR and your business system

On 14 May, Meta opened its Ray-Ban Display glasses to developers along two paths. A native mobile SDK for iOS and Android that extends an existing app onto the in-lens display, and web apps written in ordinary HTML, CSS and JavaScript, built in a browser and delivered to the glasses through a URL.

The second fact matters more around here. You cannot buy the things.

Why they are not your problem yet

In January, Meta paused the rollout to the UK, France, Italy and Canada, citing unprecedented demand and limited inventory. The $799 price including the Neural Band wristband is still a US figure and nothing else.

European regulation is the other half. From 18 February 2027, portable batteries in devices sold in the EU have to be removable and replaceable by the user. The regulation carries derogations, but a pair of glasses with a glued-in battery and a waveguide display does not fit under them comfortably.

For decision-making purposes, that is the whole story. You are not building for glasses this year. Anyone selling you that as a product is selling you a demo.

A display that fits one sentence

A lens display gives you one line and about three seconds. No keyboard, and a wrist gesture instead of a tap.

Nobody wants an application in there. What fits is a shelf number, the next customer's name, or the torque that bolt is supposed to get. Companies doing anything sensible with these devices are not using them to work inside a system. They are using them so a system can tell one thing to someone whose hands are full.

That one sentence is delivered as a web page, which is the part worth noticing.

Every new device category ships a browser

Samsung shipped Galaxy XR in October 2025 at $1,799 as the first Android XR device, and it supports OpenXR, Unity and WebXR, so anything built on the web runs there without a rewrite. Meta's glasses have their web path. At IFA in Berlin, RayNeo showed the 33-gram iO with a heads-up display, and the GT Max, which plays Dolby Vision when paired with the RayNeo Pocket TV Pro.

Different maker, operating system, input and price. They have almost nothing in common except one thing. Every one of them can open a page.

That is not a coincidence. A company launching a new device category has no app ecosystem and nowhere to get one quickly, so it ships a browser. Android XR and Meta have landed where Tesla landed years ago, on a browser nobody asked for.

Which leaves you one practical consequence that has nothing to do with eyewear.

The decision that is actually being made

What is being settled here is where the logic lives, not which screens you end up supporting.

If order status is decided inside a screen, it exists only inside that screen. Moving that fact anywhere else means working out what finished means all over again, and it will be worked out again every single time.

If order status comes back from an endpoint, it exists independently of whoever is asking. A new screen is then a day of work. A card on a pair of glasses, a watch complication, a line in Teams, a page for a warehouse worker on a five-year-old Android.

Shopify described exactly this when it split business logic away from the interface so it could run on its own and agents could query it from a command line. We went through it in their move back to native apps. The same split that lets you test the logic without a screen is what lets a second screen read it.

Who this is already worth something to

A warehouse where someone is picking and has no spare hand for a phone. A field engineer at a machine needs the next step rather than the manual. And maintenance work, where the whole gain is not having to walk back to a terminal. Two occupied hands, one short piece of information.

For office work it makes no sense. A monitor beats it on every count and will still beat it in five years.

What to do with this in 2026

This is usually the first thing we go through in an IT consultation, and it costs nothing to try on your own first.

Pick one fact your system knows that somebody in the company needs quickly. Job status, the next task in the queue, free space in the warehouse.

Expose it as an endpoint that returns one sentence. Nothing more. Then display it once somewhere other than the main application, say as a plain page on a phone next to a machine.

If that works, you own a system that outlives devices. If it does not, you have just found the most expensive feature of your software, with no eyewear involved.

Whether a separate app is worth it at all is something we took apart elsewhere, and the answer is no more often than yes.