Skip to content
  1. Home
  2. Fishbone Talk

opinion · · 6 min read

the future of qr codes: from links to buttons

Today a QR code points to a web page. With AI glasses that read what you look at, it starts acting like a button. Here's how to get your codes ready.

By Big Bad Agency

You sit down at a café. There’s a little table tent by the salt with a QR code on it. You don’t reach for your phone. You’re wearing glasses with a camera and an assistant behind it, and before you’ve thought about it, the menu is just there: today’s specials, what’s sold out, a button to order.

You never “scanned” anything. You looked at the table.

That’s not how it works today. But the pieces are closer than most people realise, and we think it changes what a printed QR code is for.

what a qr code is right now

Today, a QR code is a pointer. It holds a short string, usually a web address. You open your camera, point it at the square, tap the link, wait for a page to load, and then you get to the thing you wanted.

Every step in that chain is a place people give up. Phone in the bag. Bad signal. A page built for a laptop that takes forever on mobile. A cookie banner over the menu. The code did its job. The trip after it is what loses people.

what the cameras can already do

Here’s what’s real today, checked against the makers’ own pages.

  • Meta’s AI glasses can already read a QR code by voice. Meta’s help page for its AI glasses lists “Hey Meta, scan this QR code” as one of the things you can ask. It works by sending a photo to Meta’s cloud for processing, and the answer comes back as audio, with the request and image saved to the Meta AI app. (Meta help: Ask Meta AI about what you see)
  • Meta’s glasses don’t need to be told to look any more. When Meta announced live video for Meta AI in September 2024, it also said you “no longer need to say ‘look and’” to ask about what you’re looking at. (Meta newsroom)
  • There are glasses with a screen in the lens now. Meta Ray-Ban Display puts a small display in the right lens, and Meta says it shows “instant visual cues or answers whenever you perform an action or ask a question.” (Meta: AI glasses with display)
  • Google is building the same shape. Google’s Android XR glasses post says pairing the glasses with Gemini “means they see and hear what you do,” with “an optional in-lens display.” (Google blog)
  • Apple’s headset can find codes in the room and put content next to them. Apple’s visionOS documentation describes ARKit detecting barcodes, QR codes included, locating them in a person’s surroundings, decoding them, and placing content near them. For now it’s an enterprise API that needs a special entitlement, and its example is a warehouse worker, not a café. (Apple developer docs)

None of that is the café scene yet. Today you still ask out loud, and the answer still mostly comes back as audio or in an app. But the parts exist: a camera that’s always on your face, an assistant that reads what it sees, and a screen in the lens to show the result.

where we think this is heading

Here’s the prediction, and it is a prediction.

We think QR codes go from being links to being buttons. The device reads the code without you asking, follows it in the background, and shows you a short summary of what’s there, right where the code is. The menu floats above the table tent. The poster shows the ticket price and a “get tickets” button. The code on a shelf shows the review score and whether your size is in stock.

You stop visiting the page. You see a small preview of it, and you tap if you want more.

If that happens, the page behind your code matters more, not less. The assistant has to read it, understand it and boil it down into a few lines. A page it can’t make sense of turns into a vague, unhelpful summary, or nothing at all.

We don’t know when this happens, or which company gets there first. We’d be lying if we put a year on it.

the nft comparison (carefully)

There’s a useful idea in here that people last heard about during the NFT craze, and it’s worth pulling out without the baggage.

The good part of an NFT was never the speculation. It was a simple structure: a token that stays the same, pointing at something that can change. The token is fixed. What it points to can be updated.

A dynamic QR code already works like that, and it has nothing to do with crypto. The square you print never changes. It points at a short address you control, and you decide what’s behind it this week: the summer menu, the winter menu, a sold-out notice, a new booking link.

Put that next to the glasses future and it gets interesting. If people’s devices start showing what’s behind a code, then the printed square is a small window you own, stuck to a real place, that you can refresh whenever you like. No reprint, no wallet, no gas fees.

That’s the comparison. We’re not saying QR codes are the new NFTs, and we wouldn’t sell you either on hype.

what to do now so your codes are ready

None of this needs the future to arrive. Everything below makes your codes work better on a plain phone today.

  1. Never print a code you can’t re-point. A static code bakes the web address into the pattern for good. A dynamic code points at a short address you can change later. If the page moves, the code on ten thousand flyers still works. CreateQRCodes, our QR tool, makes dynamic codes on its paid plans, and you can change where they go after they’re printed. (We wrote up the difference in static vs dynamic QR codes.)
  2. Make the destination fast on a phone. Whether it’s a person or an assistant arriving, a slow, heavy page loses them. One clear page beats a full website with a menu buried three taps deep.
  3. Make the page readable without running JavaScript. If your page is an empty shell until scripts load, some crawlers and bots give up before they see anything. We wrote about how to tell if your prerender actually worked, and the same test applies to anything you want a machine to summarise.
  4. Add structured data. Google describes structured data as “a standardized format for providing information about a page and classifying the page content.” (Google Search Central) Marking up your menu, hours, prices or event times means a machine doesn’t have to guess.
  5. Give the page a clear title, a plain first sentence and a real image. If an assistant has a few lines to show, it’ll take them from the top of your page. “Lunch menu, Tuesday to Saturday, 11 to 3” is what you want it to show. “Welcome to our world of flavour” isn’t.
  6. Keep the facts right. A preview that says you’re open when you’re closed is worse than no preview. If you run WordPress, our free Store Hours plugin shows whether you’re open right now from one weekly schedule, so the page the code points at isn’t guessing either.
  7. Track your scans. You can’t improve a code you can’t measure. CreateQRCodes’ Pro plan includes scan analytics: totals, unique scans, country and device. (what’s in the CreateQRCodes app)

the honest part

Today most scans still happen on a phone, with a person tapping a link. Smart glasses are early, and the scene at the top of this post isn’t what happens yet. If you’re printing codes this month, build for the phone.

The good news is that you don’t have to choose. Every item on that list helps the phone scan you get today, and it’s also what a camera-plus-assistant would need later. A re-pointable code, a fast page that makes sense without scripts, and accurate facts at the top.

The square on the table won’t change. What people see when they look at it might, a lot.

Print codes you can change. Point them at pages a machine can read.

Want a site that’s built this way from the start? See what we’ve built for other businesses, or read why we’d rather build a free tool than buy another ad.

Disagree? Good.

Write to a person: hello@bigbad.agency. We answer within a business day.

Start a project