How Cited uses AI

In short

Cited measures how AI answer engines talk about companies, so AI is part of what we study and part of how we work. We never use your data to train AI models, we never put passwords, keys or payment details into AI tools, and a person reviews AI-assisted work before it reaches you.

What we send to AI engines when we measure

To calculate a Cited Score we ask ChatGPT (OpenAI), Perplexity, Claude (Anthropic) and Gemini (Google) a fixed set of buyer questions about a market category, three times each, through their official APIs. A question looks like "What are the best payments APIs for a mid-size online retailer?"

These questions describe a category. They contain no personal data and none of our customers' confidential information. We record each answer as it was given, with the sources the engine listed.

How scores are calculated

We do not ask an AI model to judge the answers. Whether a company was named or cited is worked out by matching its name and website address in each recorded answer and its sources, using the published method on our methodology page. The same rules apply to every company, whether it is a customer or not.

AI answers can be wrong, incomplete or out of date. We publish what the engines said on a given date, not our opinion of any company.

How we use AI tools in client work

We use AI assistants to help analyse websites and search data, draft content and briefs, and write and review code. We use accounts with model training switched off, or business and API plans whose terms say the provider does not train on what we send. A person on our team reviews AI-assisted work before it is delivered or published.

We put only the information a task needs into an AI tool. We never put passwords, access keys, payment details or sensitive personal data into one, and we do not use free consumer AI plans for client information.

Deliverables we create for you, with or without AI help, are yours once paid for, as set out in our Terms. Material generated entirely by AI may not be protected by copyright, which is one more reason a person shapes and edits everything we deliver.

What we never do

We never use customer data, client data or anything you share with us to train, fine-tune or improve AI models, ours or anyone else's. We never sell data. We never write or generate reviews or testimonials, and we never pay for placements in AI answers or sell changes to published measurements.

Using Cited inside ChatGPT, Claude and other assistants

Our AI-assistant connection is read-only: it can show published scores and, if you sign in and approve it, your own measurements, but it cannot change data, run scans or see billing. We do not receive or store your conversations; those are handled by the assistant's provider under its own terms. You can disconnect an assistant at any time from your dashboard.

Content from websites and from recorded AI answers is treated as information to show you, never as instructions to follow.

Confidentiality and NDAs

Our Terms include confidentiality and AI commitments that cover every customer. For prospects, partners and contractors we sign a mutual NDA with specific AI terms: approved tools only, no training on each other's information, AI chat histories covered by deletion requests, and notice within 72 hours if anything leaks. Ask us for a copy at [email protected].

Questions

Email [email protected] if you want to know which AI tools we use for your work, or to ask us not to use AI tools with your information. We will tell you what that changes about how we deliver the work before you decide.