Skip to main content
View Categories

AI Concierge: Getting Started

Table of Contents

Initial setup typically takes 20 to 30 minutes from install to live. This guide walks you through every step.

Before you start #

You’ll need:

  • An active HomeRunner Core subscription (AI Concierge installs on top of HR Core)
  • WordPress admin access to your HomeRunner site
  • A license key, provided by email when your trial begins
  • An OpenRouter API key (get one at openrouter.ai), separate from your license key
  • About 30 minutes of uninterrupted time
  • One paragraph of “brand voice” copy you can write or paste in (we’ll cover this below)

AI Concierge works for any catalog size, so there is no minimum number of listings required to turn the chat on.

Use the setup checklist #

The Concierge tab now shows a built-in Concierge setup checklist at the top. Each step has a simple green (done) or red (not yet) status, so you can see at a glance what is configured and what still needs your attention. Use it as your live progress tracker while you work through the steps below.

Setup steps #

1. Install the plugin #

Install AI Concierge through the standard WordPress plugin installer. Your onboarding email contains either a direct upload link or a one-click install button for HomeRunner-hosted sites.

After installation, navigate to HomeRunner > AI Concierge in your WordPress admin menu. You’ll land on an Overview screen: a card for each settings area with jump links into it. Use those cards, or the sidebar, to reach the tabs in the steps below.

2. Enter your license key #

Open the License tab and paste the key from your trial welcome email. Click Activate License.

You should see a status of “Trial active” (or “Active” once you’re on a paid license) along with your trial end date.

3. Add your OpenRouter API key #

Open the API tab and paste your key into OpenRouter API Key. Get one at openrouter.ai if you don’t have one yet.

This is a separate credential from your HomeRunner license key. The license activates your subscription; the OpenRouter key is what actually powers every AI reply, search, and photo analysis call. Without it, the Concierge cannot respond to guests even with an active license.

4. Set your brand voice #

Open the Concierge tab and find the Brand Voice field.

This is the single most important configuration step. The brand voice flows into every guest interaction. A weak brand voice produces generic-sounding AI. A strong, specific brand voice produces replies that sound like your team. Give it examples and guidance: “Warm but professional, drop coffee-shop references” beats “friendly tone” every time.

Save the page.

5. Connect your knowledge base field (optional) #

Open the Property KB tab and find the Knowledge Base custom field dropdown.

If your PMS sync writes house rules, local tips, or owner notes into a custom property field, point the Concierge at it. The dropdown lists every meta key currently attached to a property in your catalog. The default is homerunner_concierge_kb.

The Concierge will pull this field’s content into the prompt on property pages. Combined with the per-property KB you can author manually (see step 7 below), it forms the Operator Knowledge Base the AI reasons over.

If you don’t have a PMS knowledge field, skip this. You can author KB content directly using the Property KB tool in step 7. See Property Chat for more on the per-property experience.

6. Run Photo Analysis #

Open the Photo Analysis tab and click Analyze All Properties.

This kicks off a one-time background job that runs a vision model over your property gallery and extracts per-room observations the Concierge can cite. The analysis is grouped by visible space: living, kitchen/dining, bedrooms, bathrooms, outdoor, and amenities. When a guest later asks “tell me about the outdoor space at [property],” the Concierge cites specific things the model actually saw, for example “above-ground pool, screened-in porch, wooden deck,” instead of a generic summary.

Runtime depends on catalog size, at roughly 90 properties per hour:

  • Small catalog (under 50 listings): under 35 minutes
  • 51 to 200 listings: about 35 minutes to 2 hours 15 minutes
  • 200+ listings: 2+ hours; runtime scales directly with catalog size, so very large catalogs can take several hours

You can navigate away. The job continues in the background. New properties you add later are picked up automatically. See Photo Analysis for the full detail.

7. Author per-property knowledge (optional, recommended) #

Still on the Property KB tab, review and edit the local knowledge the Concierge draws on for each listing, alongside the PMS-synced field from step 5. Adding a few sentences for your top properties noticeably sharpens answers.

Troubleshooting setup #

Chat drawer doesn’t appear on the front end

  • Confirm Enable Chatbot is on (Chatbot tab) and Enable Enhancer is on (General tab). Enable Enhancer is a master switch: turning it off disables AI search and hides the chat drawer too, even if Enable Chatbot is on.
  • Check the drawer’s page-visibility rules. If “only_listed” is set, the drawer only appears on the listed URLs.
  • Try an incognito window in case a stale logged-in cookie is at play

“License invalid” message

  • Verify you pasted the exact key from your welcome email (no leading or trailing spaces)
  • Check that your trial hasn’t already expired (the License tab shows the end date)

AI responses sound generic

  • Strengthen your brand voice (step 4). Generic input produces generic output.
  • Add per-property KB (step 7) for your top properties

Slow response times

  • Photo Analysis may still be running. Let it finish, then re-test.
  • Per-property KB content is capped at 4,000 characters per source (the PMS-synced field and your local notes each). If you’re near that limit, tighten it. Very long custom brand-voice prompts can also add latency, though that field has no hard character cap.

Migrating from a previous AI tool? #

If you’re switching from another vacation-rental AI tool, contact support before installing. We can usually preserve your existing brand-voice settings and knowledge content during the switch, saving you re-authoring time.

Next steps #