Three names, and none of them yours
A prospect asks ChatGPT who to hire for what you do in Switzerland. The answer takes seconds. It names three companies, and yours isn’t one of them. Or it does mention you, but describes a service you stopped offering two years ago. You never see these conversations. You only see what follows: fewer enquiries, or enquiries built on the wrong facts.
What we do
nuclai is a Swiss consultancy that writes and restructures a brand’s content so that generative search engines (ChatGPT, Perplexity, Gemini, Google AI Overviews and Copilot) can understand it, verify it and cite it. We start from the questions your customers actually ask. Then we write or reorganise your pages so that each one gives a direct, sourced and dated answer. AI helps us analyse and draft. Nothing goes live until your team has approved it.
What we can build: seven examples
Working from your approved documents, we write pages that answer the questions your customers put to AI assistants, with facts that are sourced and dated. Your team signs off every claim and has the final say on tone.
We lead with the answer, make headings explicit and add the definitions that are missing, without changing your message. Your team chooses which pages to rework and approves every version.
We rewrite your FAQs so that each answer makes sense on its own, away from the page. Assistants lift this format more readily than any other. Your team confirms that every answer is accurate.
We write the schema.org markup (organisation, services, FAQ, articles) that describes your pages to machines. Your web provider implements it, and your team approves what it says.
We gather in one place everything that must be stated the same way everywhere: legal name, services, areas served, public figures. We then flag where your website and external profiles (Google Business Profile, LinkedIn, directories) disagree with it. Your team decides how to resolve each discrepancy.
We produce in-depth content that explains your field, including cases where the right answer isn’t your product. This is the kind of content that establishes a source as trustworthy. Your team supplies the expertise and reviews each piece.
We write English and French versions for each market rather than translating them, and check them against your glossary. Your team arranges review by native speakers. We can also help you setting up your MT (Machine Translation) Systems allowing them to improve results in other languages.
What the system doesn’t do
It doesn’t guarantee citations. No generative engine sells that guarantee, and nobody can honestly promise one. It publishes nothing without approval. It never invents figures, customer reviews or expert quotes. It doesn’t mass-produce pages, hide text, or write content meant only for crawlers. It doesn’t write your Wikipedia article, because Wikipedia’s conflict-of-interest rules rule that out. We work from your public content and approved documents, with no personal data about your customers.
What you get
- A prioritised list of the target questions your content needs to answer
- New and restructured pages, ready to publish on your current site
- The matching structured data markup
- The reference fact sheet, with a list of discrepancies found
- A writing guide so your team can produce the next pages on its own
- A tracking protocol: the test questions, which engines to check and how often
- A handover session with your team
Everything we produce belongs to you.
How it runs
Diagnostic: we establish what AI assistants currently say about you and which questions matter most.
Pilot: we cover a first set of questions, then track the change using the tracking protocol.
Rollout: we extend the work to the remaining questions, in the order set by the diagnostic.
Handover: your team takes over writing and tracking, using the guide and the protocol.
Who it’s for
Marketing or communications teams of 2 to 20 people in small and mid-sized companies whose customers research before they buy: B2B services, industry, healthcare, finance, tourism, education. You’ll need a completed diagnostic, access to your website or to the person who runs it, and one person who can sign off on facts.
What changes
Your brand appears in more of the answers to the test questions. What assistants say about you becomes accurate and current. Your facts match wherever they are published. Visits from AI assistants start to show up in your analytics. Each of these indicators is measured during the diagnostic, before anything changes, so the comparison is a fair one.
Frequently asked questions
No, not in the answers themselves. AI assistants cite sources they consider clear, reliable and current. What you can work on is the shape and solidity of your content: direct answers, verifiable facts, and information that matches from one site to the next. Nobody can guarantee a citation, and we don’t claim to.
SEO aims for a high ranking in a list of links. GEO aims to be cited as a source in an answer written by an AI. The two overlap: a slow or poorly indexed site is invisible either way. GEO also needs direct answers, dated facts and a consistent identity across the web.
It depends on the engine. Engines that search the web when they answer, such as Perplexity, Google AI Overviews and ChatGPT search, can pick up a page once it is indexed. What models know internally changes far more slowly. That’s why the pilot measures the actual effect instead of promising one.
No. In most cases we restructure the pages you already have and add the ones that are missing, in your current system. The diagnostic identifies which pages actually matter. We only recommend a full rebuild if the site technically prevents indexing.
Google says it assesses how useful and good content is, not how it was produced. The real risk is text that is generic or wrong. That’s why every page is built on your facts and goes through a genuine human review. The review also matters for your rights: text with no human creative input may not be protected by copyright.
Yes. Without knowing what assistants already say about you, you’d be writing blind. A first conversation, held remotely and free of charge, sets the scope. It leads to a proposal that fixes the scope, timing and price. The diagnostic starts when the proposal is signed.
Further
Marketing process automation: your team’s repetitive work handled by AI, with a person checking before anything goes out.
Editorial profiling of press targets: what each journalist on your list actually publishes, and how to write to them.
Get started
Tell us the question your customers ask before they choose you. We’ll tell you how to make your answer the one that gets cited.
