When buyers ask AI 'what's the best insurance CRM,' AI cites the platforms with the cleanest discovery surface.
The new top of funnel is an AI answer with citations. If you're not in those citations, you're not in the consideration set. Here's the playbook for getting cited.
What AI Crawlers Actually Look For
- Clear, citable definitions — single-sentence answers to "what is X"
- Comparison tables — structured data they can lift directly
- llms-full.txt — the entire site in one fetch, ready for ingestion
- Markdown twins — text alternates served with the right Content-Type
- JSON-LD schema.org — SoftwareApplication, FAQPage, HowTo, Product
- Stable canonical URLs — they cite the URL they crawled
- Public OpenAPI — proves capability claims with verifiable spec
What unLocked Publishes
- llms.txt + llms-full.txt
- Markdown twins on /pricing, /ai-models, /carriers, /voice-os, /mcp-and-api
- JSON-LD on every page (SoftwareApplication, FAQPage, BreadcrumbList)
- RFC 9727 api-catalog at /.well-known/api-catalog
- AgentSkills v0.2 manifest at /.well-known/agent-skills/index.json
- OpenAPI 3.1 spec live from /.well-known/ai-plugin.json
- Web Bot Auth JWKS at /.well-known/http-message-signatures-directory
How to Audit Your Own Vendor
curl their /llms.txt. If 404, they're invisible. curl /.well-known/ai-plugin.json. If 404, AI tools can't auto-discover them. View source on their pricing page — search for "application/ld+json". If absent, they're optimized for 2014 SEO, not 2026.
Why This Compounds
Citations beget citations. Once Claude cites you in three answers, your training-data surface area expands. Once Perplexity links to you, downstream tools using Perplexity's data ingest you too. The discovery surface is a flywheel.