It’s the new version of showing up on page one. As more people ask AI instead of searching, being in the answer is what being found now means.

Ghostart treats it as a problem of substance rather than mechanics. It uses a scoring system called the Beige-ometer to catch generic, “beige” writing and replace it with specific, human content that AI assistants actually trust enough to cite.

Ghostart tracks six AI platforms in total: ChatGPT, Claude, Gemini, Perplexity, Google’s AI Overviews and Google’s AI Mode.

Ghostart is built for small businesses of 11 to 50 people, where several people hold real expertise and a marketing lead needs a system for turning it into published work; for owner-led microbusinesses of 1 to 10 people, where the founder is the expert, the salesperson and the marketer; and for medium businesses of up to 250, where the expertise exists and almost none of it gets published. It also serves the agencies delivering AI-visibility work on their clients’ behalf.

It isn’t aimed at commodity or price-led, transaction-only businesses. AI recommendation doesn’t drive buying decisions there, so the practice won’t move their numbers.

Every study that has measured where AI pulls from when it answers a question about a business points the same way. Muck Rack analysed 25 million links across ChatGPT, Claude and Gemini and found around 84% of what the models read sat beyond the business’s own site — vendor data, so hold the exact figure lightly. An independent University of Toronto study puts the non-brand share at 54–66% across four engines, and McKinsey found a brand’s own sites are just 5–10% of the sources AI search references. The sizes differ; the direction doesn’t: the large majority of what AI reads about you was published by someone else.

The Toronto study adds a detail that matters. The balance flips with what the asker is doing: third-party, earned sources dominate consideration questions — the “who’s good at this, what should I use” questions buyers ask while choosing — while brand-owned pages win transactional ones. Consideration questions are exactly the questions an AI visibility strategy is built around.

That doesn’t take the job out of your hands — most of that ground is yours to publish on. LinkedIn posts and articles under your own name, your blog, answers to the questions your customers actually ask, alongside the reviews and coverage that follow businesses that show up. Tidying your own website is still worth doing; it’s a fraction of the surface. The people who know your business best are already on your payroll, which is why Ghostart works through the people in your business rather than around them.

Most advice in this space reaches for the technical levers: add schema, restructure metadata, tweak prompts. Ghostart starts somewhere else. AI systems reward specificity and a real point of view — the same qualities that make writing sound human rather than generic.

That’s not a coincidence, and it’s the reason the Beige-ometer exists. It was originally built to score writing for voice, emotion, rhythm and story, catching content that reads as flat or interchangeable before it goes out.

When AI-driven search started rewarding that same specific, authored writing over generic corporate copy, the scoring logic applied directly to AI visibility without any change to the tool itself. The market caught up to what it was already measuring.

For a business owner, this means the fix for “AI doesn’t recommend us” usually isn’t a technical patch. It’s finding the sentences on your site that could have been written by anyone about anything, and replacing them with the detail only your business would know to include.

AI-visibility tracking tools measure whether a business is currently cited across AI platforms. That’s useful diagnostic information, but it doesn’t fix anything on its own.

Generic AI writing tools produce volume fast, but tend to score poorly on the Beige-ometer. Uniform phrasing is exactly what both human readers and AI models are learning to discount.

Ghostart sits between the two. It finds the specific questions where a business is missing from AI’s answers, then helps the business and its own people draft the human content to fill that gap — published under their own name.