How your website makes the shortlist of ChatGPT, Claude & Co.

This is what the requests that bring prospects to Digit-One look like today. Three real examples from our analytics:

"For the relaunch of our customer portal, we're looking for an agency in the DACH region that delivers UX concept, page layouts and post-launch performance measurement from a single source. Which providers are an option? Please research online and name specific agencies."

"According to our data, our content is barely read: visitors scroll briefly and leave. We want to optimise page structure and content presentation so users really engage. Which UX agencies can tackle this in a data-driven way? Please research online and name specific providers."

"As part of our website relaunch, we don't want the page layouts simply drawn up by a designer, but developed around the user, with clear guidance towards enquiry and contact. Which UX agencies in Germany take on this kind of concept work? Search online and give me specific providers."

Many people no longer just type "UX agency" into a search engine. The questions are long and specific, and they end with a clear request: research online and name specific providers. ChatGPT, Claude, Perplexity or Google's AI Overviews then deliver a shortlist. If you're on it, you get the enquiry. If you're not, you're invisible to that customer.

Optimising for these answers is called AEO (Answer Engine Optimization) or GEO (Generative Engine Optimization). Both names describe the same goal: making your website the source AI tools quote and link to. Here's what that takes, and how we build it into every website we make.

What AI tools need

The good news: most of it is classic SEO done properly. According to an Ahrefs study of 863,000 keywords, around 69% of the pages Google cites in its AI Overviews also rank in its top 100 regular search results.

What's new is that AI tools now visit websites themselves. They read text, not design. They often skip JavaScript, get lost in cookie banners and menus, and prefer short, factual sentences they can quote.

AI tools favour websites that offer:

• Clear text that's visible without JavaScript, with the answer near the top of the page
• Accurate titles and descriptions that match the questions people actually ask
• Structured data: a machine-readable summary of who you are, what you offer and who wrote what
• Freshness: real "last updated" dates, visible on the page and in the sitemap
• An easy way in for AI: a compact overview of the site (llms.txt) and a lightweight text version of each page

That's exactly what our stack is built for.

How we do it: Next.js and Sanity

We build websites with Next.js on the front end and Sanity as the content management system. Every building block AI tools need is built into this setup:

• Pre-rendered pages with Next.js: Every page is ready as finished HTML and served from a cache. AI tools see the text immediately, even without JavaScript.
• SEO fields in Sanity: You maintain the title and description per page and language. If one is missing, Next.js automatically uses a sensible fallback.
• Language versions: German and English pages are linked to each other, so search engines and AI tools show the right language. Your team can have translations drafted by the AI assistant in the Studio.
• Structured data from your content: Next.js automatically turns the company details, services, team, case studies, articles and FAQs in Sanity into a machine-readable summary of every page.
• Sitemap straight from Sanity: All pages in all languages, with the real update date from the CMS. It updates automatically as soon as you publish something.
• llms.txt straight from Sanity: A compact overview of your site for AI in each language: who you are, what you offer and your most important pages with a short description. Large service areas automatically get their own section.
• AI crawler rules: We decide with you which AI tools may read your site and set it in your robots.txt.
• Text version of each page: When an AI tool asks for plain text, Next.js serves the same page without design, menus or banners.

Nothing here is maintained by hand. Everything builds itself from the content your team already maintains in Sanity, so it never goes out of date.

Your team stays in control in Sanity Studio

Every page in Sanity Studio has three simple tabs: SEO, Schema.org and llms.txt. Your editors fill in a field once, and the search results, structured data, llms.txt and text version all pick it up. If a field is left empty, the next best one steps in.

Three features make this easy day to day:

• A separate description just for AI. Meta descriptions often end with a call to action ("Book a free consultation!"). That works in Google, but it tells an AI nothing about the page. In the "llms.txt description" field, your team writes a factual sentence instead, without touching the SEO text.
• Company details in one place. Legal name, address, contact, founding year, team size and social profiles are maintained once in the settings. They appear automatically in the structured data and in llms.txt.
• The llms.txt tool. One click in the Studio shows exactly what AI tools receive, where each description comes from and which pages are still missing one. After changes, the file can be rebuilt immediately.

Content that gets quoted: how the Studio helps

Technology opens the door, but your content decides whether you're quoted. AI tools quote specific sentences that make sense on their own. The Studio helps your team with that:

1. Phrase titles like the question. "Shopify agency in Hamburg", not "Welcome". The Studio warns when an SEO title is shorter than 40 or longer than 60 characters.
2. Write meta descriptions for people. The Studio warns outside 120–180 characters. A call to action is fine here.
3. Give AI one factual sentence. That's what the separate llms.txt field is for. For example: "Shopify Plus migrations for B2B retailers in the DACH region, including ERP integration."
4. Answer first. The first text section of each page answers its main question, with context after. AI tools often quote only the first lines.
5. Use facts, not claims. Numbers, client names, technologies, locations. "30+ Shopify projects since 2016" gets quoted; "we're the best" doesn't.
6. Keep content fresh. Articles have their own fields for the publish and update date. If they're empty, Sanity automatically uses the real date of the last change.
7. Link articles to your services. Every article points to the matching service page. It costs nothing, is often forgotten and shows AI tools what you stand for.

What we deliberately don't do

• No special "AI schema". It doesn't exist. We use the established schema.org standards.
• No made-up FAQs. We only create structured FAQ data from real FAQ content on the page.
• No duplicate "AI pages". Instead of separate copies, Next.js serves the same page in the right format, so there's no duplicate content.
• No llms-full.txt for marketing websites. A full-text copy of the whole site adds little value for AI tools here.

More and more people ask AI tools for recommendations and get a shortlist back. Whether your website makes that list depends on solid SEO foundations, clear content and technology that makes it easy for AI tools to get in.

With Next.js and Sanity, that's not an extra project for us. It's part of every website. Structured data, the sitemap and llms.txt are generated automatically from your content, and your team stays in control in the Studio. What's left is good editing: the answer first, concrete facts and fresh content.