AI search is changing how renters find apartments. Is your property visible?

Ask ChatGPT, Claude, Gemini or Perplexity for a one-bedroom under $2,000 near the park in Montclair with in-unit laundry, and the answer comes back as a short list of buildings by name. Renters put questions to these tools the way they would put them to a leasing agent: two-bedrooms under $2,500 on the Lower East Side available in August, pet-friendly near PATH with a gym and parking, a straight comparison between your community and the one down the road. These are shortlisting questions, and they get answered out of whatever the assistant can read.

That puts a property running Engrain SightMap® in an odd position. The best facts the building owns—which units are free, what they cost, which floor they sit on—are already correct, because someone in the leasing office keeps them correct in the PMS and the feed carries them through to the map. But the standard SightMap embed is an iframe, which reads to a crawler as a blank rectangle with those facts inside it, out of reach.

What an assistant can read on your page

An assistant asked for apartment recommendations assembles its answer out of what it learned in training and out of pages it can fetch while the person waits. It reads those pages as text and markup, which leaves pictures, PDFs and interactive widgets closed to it.

Making a page legible to those systems goes by the name GEO, for Generative Engine Optimization, and structured data is the steady part of the work. A block of JSON-LD in the page source, written in the Schema.org vocabulary, states the facts outright: this is an apartment complex, this is a floor plan, these are the beds and baths, this is the rent, this is whether anything is available. Google has read this markup for years to build its rich results, and answer engines lean on the same signals. Alongside it sits llms.txt, a plain-text file at yoursite.com/llms.txt that hands a crawler an overview of what the site is and where the detail lives, the inverse of robots.txt and its list of things to ignore. It is a still-emerging convention rather than an official standard, and no model is obliged to read it, which is why we publish one without claiming much for it.

Our checklist for getting an apartment website cited by ChatGPT and Claude runs through the rest, from robots.txt and FAQ schema to freshness and one set of numbers on every surface.

What a WP FloorMap site publishes

WP FloorMap is a complete property website, running on your own Engrain subscription and your own PMS feed. The map itself is still the SightMap embed in its iframe, while WP FloorMap adds the readable layer around it, with floor-plan cards, filters and pricing rendered on the server as HTML. A renter still sees where unit 4B sits, and on which floor. The explorer arrives as a single block of map, unit cards and filters, which a marketing manager places like any other, with a floor-plan grid, a search form and a tour button available separately.

Underneath it, the JSON-LD graph is an ApartmentComplex whose nested entities are the floor plans themselves. Each is an Apartment node with a name, bedroom and bathroom counts, a floor size and an Offer carrying the price. Plans whose units are not all priced alike get an AggregateOffer instead, with a low price and a high one. Either way the node is flagged in stock or out. Amenities are published once, on the property. The unit-level detail goes into the plain-text pair: llms.txt carries the summary, the address, the unit mix, the amenities, the monthly fees and the number of units available. llms-full.txt carries the tables, floor plan by floor plan and then unit by unit.

| Unit | Floor Plan | Floor | Beds | Baths | sq ft | Price  | Available    |
|------|------------|-------|------|-------|-------|--------|--------------|
| 4B   | The Hudson | 4     | 2    | 2     | 1,100 | $2,450 | Now          |
| 7A   | The Hudson | 7     | 2    | 2     | 1,140 | $2,610 | June 1, 2026 |

Where a property publishes all-in pricing, the figures in those tables are all-in monthly totals, and the file says so in as many words. Each required fee is marked as already included, so an assistant does not add it a second time.

The graph and the text files are rebuilt from the same feed the page renders from, and every price in them resolves through the same logic as the cards, so the number a crawler reads is the number a renter sees. They update in real time without anyone retyping a rent. All of it ships with WP FloorMap, so it is there on every plan.

AI-assisted apartment search is early, and nobody outside the model vendors can say what it will reward a year from now. You can check the artifact itself. Open the file at yoursite.com/llms.txt or read the graph in the page source, both built from the feed the leasing office already maintains.