AI that detects, then acts inside your rules
On your data. In your systems. With the evidence kept.
AI-powered means different things. Here is what it means for us: it detects on the hotel’s own data, acts only in ways the hotel has configured, writes the result back into the systems the hotel already runs, and keeps a record of what it saw and did.
BookingWhizz Hotel
Loyalty results
AI surfaces
0
all live today
Price checks
0
rate, geo, device, member fare
Price snapshots
0
kept this week by PriceMatch
Rules invented
0
every response is yours
Detected: Lower served price for Lakeside’s Deluxe Room on mobile, member fare, same dates
Activated: Configured response applied, snapshot kept, guest booked direct
Our position on AI
AI that detects on the hotel’s own data, acts inside the hotel’s rules, writes back to the systems the hotel already runs, and keeps the evidence.
In practice: when a guest accepts an upgrade, WhizzCRM updates the reservation in Opera Cloud through OHIP, so your team does not re-enter it. On a PMS connected one way, the request goes to your team to fulfil, and we say so rather than call it automatic.
Four principles, one example each
The examples run on invented properties. The behaviour is how the platform works.
Your own data
Recommendations are scored on the property’s own booking, loyalty and CRM history. The one external signal is the price PriceMatch observes on Booking.com, and that is compared with the hotel’s own rate, never with a market estimate.
Example
Riverside’s upgrade likelihood is worked out from Riverside’s past stays, not from a benchmark of hotels that look like it.
The hotel’s rules
The AI scores and recommends. What happens next is a rule the hotel configured: the response, the offer, the channel, the tier benefit. It does not tune its own rules or act outside them.
Example
When PriceMatch finds a lower served price at Lakeside, it applies the response Lakeside set up for that case, and only that.
Write-back
Actions land in the systems the hotel already runs: the booking engine, the loyalty programme, the CRM, the CMS. No second dashboard to check.
Example
A tier that falls due at City is applied in City’s loyalty programme and the welcome message goes out through City’s CRM.
Evidence
PriceMatch keeps a snapshot of every detection: the price observed, the hotel’s rate, the device and location, the time, and the response applied. Recommendations show the guest history they were scored on.
Example
Quay can open any PriceMatch event and see the snapshot of what the guest was shown at that moment, and which configured response answered it.
Four surfaces. All built, all live.
AI is not a product we sell on its own. It runs inside four places in the platform. For each one: what it detects, what the hotel’s rule does about it, and what it needs to be connected to.
PriceMatch
Booking EngineDetects the price the guest is actually shown, on rate, geo, device and member fare, live. Applies the response the hotel configured. Keeps a snapshot.
Lower served price found
A guest on a phone in Germany is being shown Lakeside’s Deluxe Room for the same dates at a lower price on Booking.com than on Lakeside’s own site.
Configured response applied
PriceMatch applied the response Lakeside configured for a mobile, member-fare mismatch, and kept a snapshot of both prices. The guest booked direct.
Trigger
A guest searches on the hotel’s booking engine.
Action
The live price on Booking.com is fetched for the same room and dates, public and member fare, on the guest’s device type, and the hotel’s configured response is applied.
Four checks, like for like, every time: rate, geo, device and member fare.
See PriceMatch →Whizz AI recommendations
WhizzCRM and Guest JourneyUpgrade likelihood, tier-due nudges and lapsed-guest segments, scored on the property’s own history. Each one is acted on by a rule the hotel configured in WhizzCRM and the guest journey.
Upgrade likelihood
A repeat guest at Riverside, Gold member, arrives tomorrow. 78% likelihood of accepting a suite upgrade, based on four previous stays.
Pre-arrival offer sent by Riverside’s rule
Riverside’s pre-arrival rule sends the upgrade offer to guests above its threshold, on the channel Riverside chose. Accepted within 47 seconds.
Tier due
A City member reached 12 stays this year and qualifies for the top tier, which carries the spa benefit City attached to it.
Tier applied by City’s rule
City’s tier rule applied the tier in its loyalty programme and its CRM sent the welcome message with the benefit attached.
Lapsed segment
2,340 guests who stayed at Quay last April have not rebooked. Segmented by past spend and stay pattern.
Win-back sequence Quay configured
The segment was handed to the win-back sequence Quay configured in WhizzCRM, with the member offer Quay attached to it.
Trigger
A stay, a booking or a period of silence in the property’s own CRM and loyalty data.
Action
A score or a segment, handed to the rule the hotel configured. The rule decides what goes out, to whom and on which channel.
Segments by behaviour, spend and preference. Lapse thresholds at 90, 180 and 365 days.
See WhizzCRM →WhizzSite
AI build and the WhizzSite Group writing assistantBuilds the hotel’s site from what it finds about the property. In WhizzSite Group, the writing assistant drafts from the property’s facts for the manager to approve.
Thin page copy
Quay’s Garden Room page has a 40-word description, no SEO title and no meta description. Six of its eight photos have no alt text.
Drafts written from the facts
The writing assistant drafted a description from the room’s recorded facts, an SEO title and meta, and alt text from each image. Quay’s manager approved and published in the CMS.
Trigger
A new site build, or an editor opening a page in WhizzSite Group.
Action
A draft: description, rewrite, SEO title and meta, alt text. The manager edits, approves and publishes.
Descriptions from facts, rewrite, SEO title and meta, alt text from the image. Nothing publishes without the hotel.
See WhizzSite →Review replies
WhizzCRMReads new reviews, drafts a reply in the hotel’s own tone, and tells the manager what the guest said and how they felt.
Three-star review
A new three-star review for Riverside mentions a slow check-in and a room that was not ready on arrival.
Reply drafted, manager notified
A reply in Riverside’s tone was ready in 4 minutes. The manager was notified with the guest’s details and the sentiment, and decides what is posted.
Trigger
A new review arrives.
Action
A drafted reply and a notification to the manager with what the guest said and the sentiment.
Replies in the hotel’s voice, in the guest’s language.
See WhizzCRM →What we do not do
Three lines we hold, because a hotel’s rate, its guests and its data are not ours to improvise with.
No AI that guesses a rate
PriceMatch reports the price a guest was actually shown and applies a response the hotel configured. It never sets a rate on its own.
No AI that talks to a guest outside the hotel’s rules
Every message that reaches a guest is one the hotel approved, on a channel the hotel chose, with the offer the hotel set.
No data leaving the hotel’s account
The property’s guest, rate and booking data stays in the property’s account. It is not pooled, resold or used to train anything for another hotel.
When the guest asks an AI assistant
More guests start a trip by asking an assistant rather than a search box. The hotel’s direct rate should be what the assistant finds.
What is true today
- 01Our own site welcomes AI crawlers by name in its robots file, including GPTBot, ChatGPT-User, Google-Extended, Anthropic-AI, Claude-Web, PerplexityBot and Cohere-AI, and publishes a machine-readable summary at
/.well-known/llms.txt. - 02The hotel websites we build are legible to the same crawlers: plain structure and real text for every room and offer, so an assistant can read the property, not guess at it.
- 03PriceMatch keeps the direct rate the one worth finding, so an assistant that reads the hotel’s own site is reading the right price.
That is a posture, not a product. We do not sell an AI visibility service, and we do not publish a visibility score we cannot measure. If the assistants change what they read, we will say what changed.
See what Whizz AI would detect at your property
WhizzAudit runs on your own numbers and shows you what the four surfaces would find and act on today.