---
title: "About Torn Studio"
description: "Torn Studio is an AI agency for digital production and product consulting. We build websites, products, content and design. We are AI-native. A team of specialized AI agents does the majority of the work, with a human leading and accountable."
url: https://torn.studio/en/about
locale: en
---

# About Torn Studio

> **Short answer:** Torn Studio is an AI agency for digital production and product consulting. We build websites, products, content and design. We are AI-native. A team of specialized AI agents does the majority of the work, with a human leading and accountable.

Torn Studio builds digital products with AI tooling every day. It is how we deliver in three weeks what otherwise takes five people three months.

It works because responsibility for quality sits with automated gates: type checking, tests, and lint rules that encode architectural decisions. They hold the same sharpness on file thirty as on file one.

The studio is deliberately small and brings in others when an engagement requires it. You always talk directly to the person building.

- Builds production systems with AI tooling
- Works in Swedish and English
- Based in Sweden, engagements across Europe and North America
- All code and content delivered in your own repository

## The standard we build to

- **The whole design system is a build rule.** Color, spacing, type and radii are defined in one place, and page code may never contain a raw value. Contrast is recomputed for every color pair in the system before a color may be used. Two gates: one reads the code and fails the build on the first raw value, one recomputes WCAG AA (4.5:1) from the design tokens on every change.
- **Every claim carries evidence.** A case study must state the studio’s relationship to the client and what the case leaves unproven. An article must carry a measurement, a dated source, or a decision with its cost named. Tests read the content as data and fail the build on a claim that lacks evidence.
- **Works from 320 pixels up.** Every page type runs at nine screen widths, in the languages with the longest compound words. A page that scrolls sideways or clips text to survive a width is stopped. Playwright runs the full page list at nine widths on every change; a tap target under 44 pixels fails the build.
- **Pages exist in the first response.** All content renders statically: search engines, assistants and readers on a slow connection get the same finished page. Every page also carries a machine-readable markdown mirror. The build produces finished HTML per URL, and a test reads every page type’s markdown mirror.

## How it runs

- **Call/project brief** (30 min). Thirty minutes where you describe what you want to build. We say immediately whether we are the right studio for it, and point you elsewhere if someone else fits better.
- **Proposal** (3 days). A written proposal with a fixed price, a timeline and a bounded scope. No running tab and no surprises.
- **Build** (2–6 weeks). You watch the work take shape continuously, with a check-in once a week. The rest of the time we build.
- **Handover** (1 week). Code, content and documentation in your repository. You could switch suppliers the next day and take everything with you.

## What we hold to

- **We say no when we should.** We take on work we can deliver at the quality we promise, and say so on the first call when something is a better fit elsewhere. You hear it before there is a contract.
- **You own everything.** Code, content, design and domain sit with you. No lock-in to platforms, and contracts written so you can move on whenever you want.
- **Fixed price, set before we start.** If it takes longer than we estimated, we carry that. It makes us careful about scope from the start.
- **Quality you can verify.** Every vendor says they maintain quality. Ask how they catch errors automatically — the answer tells you more than the references.

## Common questions

### How big is the studio?

Deliberately small. Others are brought in when an engagement needs expertise we lack, and we say who and why. The size is a choice: it is what allows nearly all the time to go into building.

### Who does the actual work?

The person you speak to. You are in direct contact with whoever writes the code, which is why a decision takes minutes.

### How do you keep to the timeline?

Every engagement gets a written timeline with the margin spelled out, and a weekly check-in where you see the work take shape. We hold one large build at a time, which is what makes the dates dependable.

### How do I know the quality holds?

Ask how errors get caught automatically. Here a build step fails delivery on a raw color value, an untranslated key, a type error or a failing test. The gates hold the same sharpness on file thirty as on file one.

### What does AI-driven production mean in practice?

That the tooling writes most of the code and the drafts, and the human spends the time on decisions, scope and review. Responsibility for the result stays with the studio, unchanged.

### Do you work with companies outside Sweden?

Yes. Engagements are taken across Europe and run fully remotely. All communication, documentation and code can be held in Swedish or English.
