Squintwerks

You are standing in front of a poster. Maybe it is a movie poster. Maybe it is an ad on the side of a bus shelter. Maybe it is a piece of art on a gallery wall, or a vintage print in a café, or a sign outside a restaurant you have never been to. You look at it. You think: I wonder what that is. And then you keep walking, because there is no easy way to find out.

Squintwerks is the answer to that moment. Point your phone at the image. Watch. Fifty seconds of content — the story behind what you are looking at, delivered on the spot, right there, always on. No searching. No typing. No waiting. The world becomes watchable.

It sounds simple. It is not. Building it is one of the hardest things a small team can attempt in 2025 — and understanding why requires a brief, honest account of what it actually means to make software today.

The Difficulty of Building Software Today

There is a version of the story of software development that goes like this: the tools have never been better, the talent has never been more accessible, the cost of building has never been lower, and anyone with a good idea and a laptop can ship a product to the world. This version is not wrong. It is just incomplete.

The complete version includes the part where the tools change every eighteen months. Where the framework you chose two years ago is now deprecated. Where the AI assistant that was going to write your code for you turns out to write confident, plausible, subtly broken code that takes longer to debug than it would have taken to write correctly in the first place. Where the cloud provider that was going to make infrastructure invisible has a pricing model that requires a full-time person to understand. Where the open-source library you depend on is maintained by one person in Germany who has a day job and a new baby.

The complete version includes the part where you are not just building software. You are building software while also doing product design, user research, customer support, marketing, legal compliance, security auditing, and — if you are a small team — probably also answering your own emails. The barrier to starting has never been lower. The barrier to finishing has never felt higher.

"The hardest part of building something new is not the technology. It is staying convinced, on the days when nothing works, that the thing you are building is worth the effort."

And yet. People keep building. Because the alternative — not building, not trying, not making the thing that exists only in your head — is worse. The difficulty is real. So is the pull.

What Squintwerks Is Trying to Do

The idea behind Squintwerks is deceptively straightforward: the world is full of images that have stories behind them, and most people walk past those stories every day without knowing they exist. A movie poster is not just a movie poster. It is a production history, a cast, a director's vision, a cultural moment. A restaurant sign is not just a sign. It is a menu, a review, a neighbourhood, a recommendation. An advertisement is not just an advertisement. It is a brand, a product, a claim you might want to interrogate.

Squintwerks makes those stories accessible in the moment you are standing in front of them. Point your phone at the image. The camera recognises it. A short video plays — fifty seconds, content-dense, pertinent to exactly what you are looking at, delivered by an avatar who knows the subject. You don't leave the moment. You go deeper into it.

Think about what that means for a museum. For a gallery. For a high street full of ads that currently do nothing except exist. For a magazine — like this one — where every image could become a portal to the full story behind it. You see the Audrey Hepburn photo. You point. You watch fifty seconds on the history of Tiffany & Co. You keep walking, but you know something you did not know before.

The Technical Reality

Image recognition at the speed and accuracy required for a "point and watch" experience is not a solved problem. It is a hard problem that has been made easier by a decade of machine learning advances — but easier is not the same as easy. The system needs to recognise images reliably in variable lighting, at odd angles, partially obscured, on glossy surfaces, on matte surfaces, in direct sunlight and in dim interiors. It needs to do this fast enough that the experience feels instant, not laboured.

Then there is the content layer. Recognising the image is only the first step. Delivering the right fifty seconds of content — accurate, engaging, properly produced — requires a content pipeline that is itself a significant undertaking. Who makes the videos? How are they kept current? How do you handle an image that has not been registered in the system? How do you build a library large enough to be genuinely useful without spending years doing it?

These are the questions that make Squintwerks hard. They are also the questions that make it interesting. The answers are not obvious. The path is not straight. This is what it looks like to build something that does not exist yet.

Why It Matters

We are surrounded by information we cannot access. Not because it does not exist — it does, somewhere, in a search result or a Wikipedia article or a database — but because the friction of finding it in the moment you want it is too high. By the time you have typed the query and scrolled through the results, the moment has passed. You are somewhere else. The curiosity has cooled.

Squintwerks is a bet that the friction can be reduced to almost nothing. That the gap between "I wonder what that is" and "now I know" can be fifty seconds of well-made content, delivered the instant you point your phone. That the world — the physical, visible, walkable world — can become a library you navigate with your camera.

It is in progress. The hard parts are still hard. But the idea is right, and the people building it know it. That combination — a right idea and people who believe in it — is, historically, enough.

"Don't leave home without it. Seriously — you'll miss some awesome stuff."

— squintwerks.com