A boutique 40-room city hotel runs on booking exports, channel-manager spreadsheets and gut feel. We built the operations brain: one dashboard where the numbers arrive cleaned, explained and forecast — with the AI doing the analyst's job. This is a showcase build of our InsightAI product configured for hotel operations, and the full product is live for you to test.
↑ That's the real dashboard rendering live inside this page — every chart is computing from the demo dataset right now.
The brief came from a familiar place: a well-run boutique hotel with healthy occupancy — and no single view of the business. Reservations lived in the channel manager, revenue in accounting exports, OTA commissions in a spreadsheet nobody trusted, and the weekly "how are we doing?" answer took a manager half a day to assemble every Monday.
What they asked for, in their words: "One screen that tells us how the hotel is actually doing — without anyone building reports. And when something looks off, we want to know before the end of the month, not after."
Booking exports drop in as CSV — the cleaning agent maps the columns, normalises dates, strips duplicates and logs every fix it made. Messy channel-manager exports become one trusted dataset, no manual prep.
Revenue, bookings, guests and average booking value with period-on-period deltas, weekly revenue analytics, channel mix, and top revenue sources across rooms and services — deluxe kings to breakfast add-ons.
This is where the analyst lives: statistical anomaly alerts on every week of trading, a 12-week trend×seasonality forecast with honest confidence bands, guest RFM segmentation with suggested retention moves, and an insight feed written by the engine from the numbers themselves.
A copilot that computes answers from live data — "why did revenue spike?", "best room type?" — a what-if simulator for rate and marketing decisions, and the Monday owner report generated in one click, print-ready.
Z-score scan (σ ≥ 2.1) across every trading week — flags spikes and dips with severity before a human would notice.
Trend × seasonality model fitted to 26 weeks of history, projecting 12 weeks ahead with an 80% confidence band.
Turns computations into written findings — strongest day, channel concentration risk, fastest-rising room type.
Ask in plain language; it aggregates the live dataset and answers with numbers, not canned text.
RFM scoring sorts guests into six segments — champions to at-risk — each with a suggested retention play.
Column mapping, date normalisation, dedupe and type coercion — with a transparent log of every fix.
Rate elasticity and diminishing-return ad models let the owner test decisions before spending a dirham.
The Monday report, written by the engine: KPIs, channels, anomalies, outlook and recommended actions.
The full product is live with the hotel demo workspace loaded — click every chart, ask the copilot, import your own CSV, print the report.
Test the working product