When a guest opens the chat while looking at one specific listing, Property Chat lets the AI Concierge answer about that home, grounded in its real details, instead of treating every question as a fresh catalog search.
What it does #
On a single property page, Property Chat hands the AI Concierge a compact, structured picture of the listing the guest is viewing, so it can answer hyper-specific questions like “is this good for toddlers?”, “is there a private pool?”, “how far is downtown?”, or “can I bring my dog?” without launching a new search.
The context the assistant draws on for that property includes:
- Core facts: title, location, property type, bedrooms, bathrooms, guest capacity, and sleeping arrangements.
- Amenities and tags assigned to the listing.
- Descriptive copy already on the listing (intro, the home, neighborhood, and notes), trimmed to keep answers fast.
- Operator knowledge base (see below), treated as authoritative.
- Visual observations from Photo Analysis when that property has been analyzed.
On a property page, the Concierge gets that listing’s full amenity list, not the shortened sample it works from when comparing many homes at once in a search. That means once a guest is asking about one specific home, the assistant has the complete picture of what it offers, so it can confidently answer detailed amenity questions instead of wrongly telling a guest “I don’t see that listed” for something the home does offer.
When this property context is active, the assistant answers in conversation mode and keeps replies short and grounded. If the guest pivots to wanting a different home (another city, different dates, more bedrooms), the Concierge drops the single-property focus and runs a normal search instead.
When that broader search comes back thin or empty, the Concierge does not just say no. It offers one-tap ways forward: a chip to search beyond the page’s own area or category (“Search all areas,” or “Beyond Telluride” when a location narrowed things down), and, when loosening any single filter on its own would not have helped, a chip that relaxes two at once (“Different home type at a higher budget,” for example). That combined option only appears when it would actually surface homes.
If the assistant recommends one specific home out of a set of results, it names the exact listing so the reply reliably links to that property’s page, using the short, conversational form of the title (dropping a location suffix like “| Telluride”) rather than reading back the full PMS-style name.
The same safety rule that governs the rest of the Concierge still applies here: the assistant answers only from the details it has been given, and refuses to invent facts it cannot confirm.
What is automatic vs. operator-configured #
Be clear on which parts you control:
- Automatic: the core facts, amenities, descriptions, and any photo observations are pulled from the listing for you. You do not configure these per property.
- Operator-configured: the knowledge base is yours to set. The Concierge will not have house rules, quirks, or local tips unless you add them.
The per-property knowledge base #
The knowledge base is supplemental, authoritative knowledge you write for a specific listing: anything that is not already in the listing description. It comes from two sources, and the Concierge combines both at chat time:
- A PMS-synced field. If your PMS feeds a custom field of concierge knowledge into each property, you point the Concierge at it once and it is used everywhere automatically.
- A per-listing local field you author by hand inside WordPress.
Good knowledge-base material includes things like:
- Smart-lock or check-in instructions and timing.
- Pet policy specifics (“dogs only, refundable deposit”).
- The nearest grocery store and how long the walk is.
- The owner’s favorite nearby restaurants.
Each source is capped at 4,000 characters, with a soft warning at 3,000. Keep entries tight: long knowledge bases slow responses and dilute the assistant’s focus, so cover only what is not already in the listing description.
Where to find it #
Turn it on #
Go to HomeRunner -> AI Concierge -> Property KB and, in the Property Page Experience section, switch on Enable Property Page Chat. This master switch is off by default. While it is off you can still author knowledge-base entries, but they will not reach the chatbot, and property pages get the standard welcome rather than the property-aware one.
The same section has a companion toggle, Attention Pulse on Every Visit, which is on by default. With it on, the launcher gets a brief attention-pulse animation every time a visitor lands on a listing page with chat closed. Turn it off and the pulse becomes more restrained: it only plays once per visitor session, and only when your property tooltip’s auto-show is also turned off.
Configure the property tooltip #
The same Property Page Experience section also controls a small tooltip next to the chat tab on listing pages, inviting the guest to ask about that home. Leave the Property Page Tooltip text blank to keep the standard tab tooltip, or write your own using {property} to insert the listing’s name (for example, “Ask me anything about {property}!”).
Four more toggles shape when and how it shows:
- Auto-show Property Tooltip (on by default): shows the tooltip on its own after a short delay, once per browsing session.
- Auto-show Property Tooltip After: the delay before it auto-shows. Five seconds by default.
- Show Property Tooltip on Hover (off by default): also shows the tooltip whenever a visitor points at the chat tab.
- Tooltip Once Per Visitor (off by default): remembers a visitor’s browser so the tooltip shows only once ever, rather than once per browsing session.
Point at your PMS knowledge field #
Open HomeRunner -> AI Concierge -> Property KB. There you choose which synced custom field holds your concierge knowledge (the default field is homerunner_concierge_kb). The picker lists the custom fields discovered across your published properties, so you can select the right one rather than typing a field name from memory.
Author and review local knowledge #
Two places write to the same per-property local field:
- On any individual property, an AI Concierge: Local Knowledge box appears on the edit screen (it shows up once Property Page Experience is on). Type the listing’s notes there with a live character counter.
- In the Property KB tab, a single table lists every published property so you can see at a glance which listings have knowledge authored, scan previews, filter to the ones still empty, and edit any entry inline without opening each post.
How it sharpens answers #
With Property Chat on and a property page open, a guest gets a property-aware welcome (“Ask me anything about this home, or have me find you something else”) and answers that combine the listing’s own facts, your knowledge base, and any photo observations, all under the assistant’s no-invented-facts rule. The result is fewer “I’m not sure” deflections and more confident, on-listing answers.
Watch Management #
This depends on the Conversion Kit plugin. Watch Management only works on sites that also run Conversion Kit with its Availability Watchers tool turned on. Without it, the Concierge tells the guest plainly that availability alerts are not set up on the site yet, and offers to take their contact details instead of a watch.
A guest looking at a property (or dates) with no availability can just ask, in plain conversation, for the Concierge to keep an eye out: “let me know if this opens up for the first week of August,” for example. The Concierge asks for exact dates if none were given (it will not guess at a “sometime in August” window, since a guest is going to be emailed about this), and figures out which home or homes to watch from whatever the guest already had in view: this property page, a set of search results just shown, or their saved wishlist.
The Concierge does not create the watch itself. It replies with a watch card showing the proposed dates and home(s), and the guest confirms by entering an email address and submitting the card, then confirming a second time from a link in that email. From there, the watch itself is created and tracked by Conversion Kit’s Availability Watchers tool, not by the AI Concierge. The chat is only where the guest asks for it.
Checking and cancelling a watch #
A guest can come back later in the same chat session and ask something like “what am I watching” to get a list of the watches they set up, each with its dates, home(s), and status (waiting on email confirmation, active, or paused). Asking to cancel one works the same way in reverse: the Concierge never cancels a watch on its own. It surfaces a confirm option, and only the guest’s own click sends the cancellation.
If something goes wrong, such as a database hiccup, the Concierge says so plainly (“try again shortly”) instead of guessing, and it will never tell a guest they have no watches when it actually could not check.
Tips #
- Author knowledge for your highest-traffic listings first. The Property KB table’s “Empty” filter makes the gaps obvious.
- Lean on the PMS field if you already maintain concierge notes in your PMS so you set the source once instead of copying into every listing.
- Keep entries under 3,000 characters and reserve them for what the description does not already say.
- Availability alerts need Conversion Kit. See Conversion Kit: Availability Watchers to turn that tool on before guests can use Watch Management from chat.
- For where Property Chat sits among the other operator surfaces, see Operator Tools and the Overview.