Introduction

This page sets out how digiDen Studio ("we," "our," or "us") uses artificial intelligence in our work — specifically Claude, made by Anthropic.

We've written it plainly, because the alternative is worse. Plenty of studios use AI and say nothing. We'd rather you knew exactly how the work gets made, so you can judge it on what it actually is.

Our Position

We use Claude to write, and how much varies with the job.

Some things start as Su talking — a voice note, recorded on the move, in completely the wrong order. Claude's job there is to get it into a shape you can read. The thinking, the opinions and the turns of phrase are hers. What Claude does is help them out of her head.

Other work — website copy, proposals, client documents — Claude drafts more of, working from Su's direction, her notes and her voice. She then rewrites it, cuts it, and signs it off.

That's a deliberate choice, and we think it's a good one. Claude helps us think faster, hold more detail at once, and get further in a working day than we could otherwise. It means a small studio can give you the depth of attention a much bigger one would charge for.

What it doesn't do is decide what we think. The judgement, the direction, the opinions and the sign-off are Su's. Claude is a very good writing partner working from her material, her voice and her instructions — not a machine we've handed the studio over to.

We're not using AI to fake anything. We're using it to do better work, and we'd rather tell you that outright than have you wonder.

Where We Use AI

We use Claude, and a few other AI tools, in the following ways:

• Research and analysis — gathering information, summarising long documents, identifying patterns in data.

• Writing — website copy, proposals, onboarding documents, portal updates, client notices and internal documents. Drafted with Claude, directed and edited by Su. Su's newsletters and letters work the other way round: she dictates them, and Claude shapes what she said.

• Code assistance — generating, reviewing, or debugging code as part of a website build.

• Design support — image assets and the occasional starting point. Design decisions, layout, and creative direction are done by hand.

• Audit work — AI-assisted analysis is part of how we deliver the Site Health & AI-Readiness Audit (see those T&Cs for detail).

• Internal operations — tools that help us run the studio efficiently (note-taking, scheduling, organising).

Where the Human Line Sits

AI drafts. Su decides. In practice that means:

• Final sign-off — every deliverable that reaches you has been reviewed by Su personally.

• Strategic decisions — choices about positioning, direction, and creative judgement are made by humans.

• Client comms — a person reads your email and a person answers. Claude may help draft the reply; it never sends one on its own.

• Training AI on your content — we don't submit your private content, designs, or business information to AI tools for training purposes.

Watermarking and AI Disclosure

From August 2026, the EU AI Act requires AI providers to mark the text their models generate. Anthropic — along with around 190 other signatories to the EU Code of Practice on Transparency of AI-Generated Content — has begun doing this, and it applies globally, not just in the EU.

In practical terms:

• An invisible watermark — newer Claude models embed a statistical pattern in the words they choose. Nothing is added to the text, there are no hidden characters, and it doesn't change the quality or meaning of the writing.

• No identifying information — the watermark can't be traced back to a person, a business, or a conversation. It only indicates that Claude was likely involved.

• What it can't tell you — it can't separate writing from heavy editing, and on short passages or lightly edited human writing there may be too little for it to register at all.

• Images and files — where Claude produces a file such as a .png, .jpg or .svg, it attaches a signed note in the file's metadata (an open standard called C2PA) saying the file was made or processed with Claude. Again, nothing identifying.

• Ownership is unaffected — a watermark doesn't change who owns a piece of work or who's responsible for it.

We're mentioning this because you may start seeing AI-detection claims made about content generally, ours included. Our answer is simply that we've already told you. If a detector flags our writing, it's confirming what this page says rather than exposing anything.

Anthropic explains the method in full here: How Claude's text watermark works

Data and Privacy

When we use AI tools as part of our work:

• We don't share personally identifying client information with public AI models.

• We're selective about which AI tools we use, and we favour those with strong data protection policies.

• For sensitive content (financial information, personal data, confidential business details), we either don't use AI or use it locally in a way that doesn't share data with third parties.

• Where AI tools require account-level data to function, this is handled under our Privacy Policy and the third-party processor's own privacy terms.

AI in Our Audits

The Site Health & AI-Readiness Audit explicitly uses AI tools as part of its methodology — both to analyse your site and to assess how well it performs for AI tools (search engines, AI assistants, summarisers). This is set out in the audit T&Cs.

Your Choice

If you'd prefer we didn't use AI on your project, tell us. We'll talk through what that looks like in practice — it usually means longer timelines and a different price, because the speed is part of what we're passing on to you. Either way, we'll respect your preference and we won't be odd about it.

Updates to This Page

The AI landscape is moving fast. We'll update this page as our use of AI evolves. Significant changes will be reflected in the "Last updated" date below.

Contact

For any questions about how we use AI in our work, please contact us at create@digiden.studio.

Last updated: 30 September 2026