Atlas is the system behind steps two through five of the Visigent loop: a governed model of your business, the complete set of pages that express it, and the monitoring that keeps both current.
Every page has a reason to exist, or a reason to go.
Models build their picture of you from what they can retrieve. Most businesses publish a fraction of what they know and sell, and much of that is stale or contradicts itself. That gap, not a missing keyword, is why you get left out of the answer.
Atlas closes it. It turns everything the business knows into structured, validated facts, decides which pages should exist to express them, and keeps the two in sync as the business changes. AI does the repetitive work: extraction, comparison, drafting, checking. People make the calls.
Every product, service and plan, with what it's for. Every location and the markets it serves. The people, their credentials and what they're authorities on. Pricing, policies, logistics, what to expect. Proof: reviews, awards, case studies, certifications. The questions customers ask before they buy. Each becomes an entity with a definition, a source, an owner and its relationships to the rest.
The model is mapped against your site, page by page. Each existing page is kept, updated, consolidated or retired; each missing fact gets a page built only if it needs one. Page count is not the goal. Coverage is.
Businesses change weekly. Websites don't. When a fact changes in the model, Atlas pushes one verified change to every page, schema object and FAQ it touches, and the tracker watches whether AI's description of you changed with it.
Atlas compares what the business knows with what the website says, and classifies every gap: missing, stale, unsupported. Each classification becomes a page decision: keep, update, consolidate, build.
Missing, thin, contradictory or stale information creates uncertainty. AI systems resolve uncertainty by leaving you out.
Revelus Dermatology, Austin. Every condition, service, provider and policy has a page of its own, connected to the rest. Nothing about the practice is left for a model to infer.
| Published record, by what it describes | Covers | Pages |
|---|---|---|
| Conditions | Problems it treats | 45 |
| Cosmetic services | Elective offerings | 38 |
| Medical services | Appointments, surgery, Rx | 30 |
| Resources & policies | Insurance, forms, policies | 34 |
| Providers | People and credentials | 5 |
Page counts from the site's own sitemaps, September 2026. Page count is not the goal; coverage is. Each page has a defined reason to exist.
of the domain, 13 Jul to 6 Aug 2026. Ahrefs.
of 300 category prompts answered with Revelus, 22 Jul run.
when mentioned, across ChatGPT, Claude and Gemini.
in the same run.
Measured with the Visigent tracker and Ahrefs. Read the case study
Those retrieve internal documents for an application. Atlas is the public layer: the authoritative, crawlable record that search engines, AI assistants and agents learn your business from.
A calendar produces posts. Atlas produces a complete record and then maintains it.
Atlas does not publish to your site. Agents draft, check, compare and recommend. Your people approve and publish. The approval boundary widens only when confidence and controls justify it.
Atlas runs inside the platform from the Pro plan up, with knowledge gaps and the page-by-page plan, and in full on Agency, with templates, schema and implementation specs. On the managed program, our team builds and maintains the record for you.
Written by Tom Howell, Co-Founder & CEO. Reviewed by Mikaela Berman, VP Product Growth & CX. Updated September 2026.
Enter your website. In 24 hours: your GEO Score, how each model describes you, where you're missing, and the first three fixes.