Technical and content AI visibility.
AI visibility is not a content problem and it is not a technical problem. It is both, simultaneously. Teams that ship only one half watch their citations stall.
The two layers, side by side
AI engines pick a citation when two things are true at the same time: the model can access the fact (technical), and the fact is worth quoting (content). Miss either side and the citation goes to a competitor.
- Technical layer — JSON-LD schema, AI sitemaps, structured endpoints, agent manifests, crawlability for GPTBot / ClaudeBot / PerplexityBot, fast server response, no JS-blocked content.
- Content layer — passage-level answers to specific queries, defensible claims with sources, third-party citations from Reddit / G2 / industry press, entity consistency across the web.
What breaks when you ship only one
Content-only teams publish excellent passage-level answers, but GPTBot can't parse the page because the schema is wrong, the AI sitemap is missing, and half the body content renders client-side. Technical-only teams deploy a perfect machine-readable layer, but the model still skips them because the underlying content is generic, hedged, or unsourced.
The technical checklist
- Organization, WebPage, Product/Article/FAQ JSON-LD on every relevant page.
- An /ai-sitemap.xml separate from your human sitemap.
- Structured JSON endpoints at stable URLs (company profile, products, FAQ, case studies).
- An agent manifest at /.well-known/ai-plugin.json.
- Explicit allow rules in robots.txt for GPTBot, ClaudeBot, PerplexityBot, Google-Extended.
- Server-side rendering or prerender for the body content AI engines need to extract.
The content checklist
- One direct answer per query, in the first 60 words of the relevant passage.
- Defensible claims with named sources, dates, and numbers — not adjectives.
- Consistent entity description across your site, Wikipedia, LinkedIn, Crunchbase, and Reddit.
- Third-party citations on the sources AI engines actually trust for your category.
- Page-level passages mapped to specific AI queries, not generic topics.