Real estate · Huntington Beach
A 50-year name, rebuilt so AI can read it.
Craig Ratowsky has sold homes in Huntington Beach since 1977, more than 900 of them across two generations. The reputation was real. The website was not built for how AI decides who to name. We replaced the platform, structured the whole record for machines, and set a content engine running that publishes on its own every day.
The numbers
- 339
pages indexed
answer-first, schema on every one
- 34
JSON-LD schema types
cross-linked knowledge graph
- 14
AI crawlers allowed
GPTBot, ClaudeBot, PerplexityBot and more
- 1 / day
articles published
fully autonomous since July 2026
- 35
five-star reviews
real, attributable, structured for citation
- 900+
homes sold since 1977
the record AI could not previously see
The problem
The Ratowsky record is one of the deepest in Huntington Beach real estate. Craig has been licensed since 1977. His son Justin is a third-generation California Realtor. Together they have sold more than 900 homes, hold a 5.0 rating across 35 reviews, and rank in the top 1.5 percent nationally by RealTrends.
None of that reached AI. Their previous site ran on a template platform (Luxury Presence) that scored 39 on mobile Lighthouse, with a largest-contentful-paint of 21.9 seconds. It was slow, its links were not fully crawlable, and its content was not structured the way answer engines read expertise. When a buyer asked ChatGPT who the best agent in Huntington Beach was, the deepest name in town was not in the answer.
What we built
The build, step by step.
- 01
A rebuild on Next.js, structured for machines
We replaced the template platform with a Next.js 15 site of 339 pages. Every neighborhood, guide, and market report leads with a direct answer, then backs it with real local detail, so an answer engine can lift a citation cleanly.
- 02
A 34-type schema graph and an AI-crawler allowlist
Thirty-four schema.org types (Organization, RealEstateAgent, Person, Place, FAQPage, Review and more) are emitted with stable identifiers so entities cross-reference each other. An explicit robots policy names 14 AI crawlers to allow and blocks mass harvesters. A hand-curated llms.txt gives the models the facts up front.
- 03
An autonomous daily content engine with a compliance gate
Every day the engine selects a topic, writes an answer-first article in the Ratowsky voice using Claude Opus, and runs it through a Fair Housing and RESPA compliance gate before it publishes itself. A post that fails the gate never ships. No invented stats, no named-lender endorsements, no describing people instead of property.
- 04
A live MLS map on a real feed
An interactive map plots active listings and the sold track record on a dual Trestle feed from Cotality (CRMLS), with a public IDX view and a registration-gated VOW view. The site shows the work, not a stock illustration of it.
“Craig and Justin have been the only realtors we have used for 35+ years. They are simply the best.”
Brenda Ellis
Client since the 1980s · one of 35 verified reviews now structured for AI
The outcome
The infrastructure AI reads is now in place: 339 structured pages, a 34-type schema graph, an llms.txt the models can parse, an open door for 14 AI crawlers, and a content engine that adds a compliant, answer-first article every day without anyone touching it.
The proof that took 50 years to earn is finally legible to the machines buyers now ask first. A Trinidad Island waterfront listing sold for $3,925,000, about $643,000 over asking with 12 offers in 8 days, and that kind of result now lives on a page an answer engine can cite by name.
Want this for your name?
Start with the free visibility check, or go straight to your build.