AI workforce management for hotels on Shiji.
HotelCadence is a workforce management platform for hotels: it reads the reservation data already in Shiji Enterprise Platform (via its API, read-only) and turns it into hourly demand forecasts per department and auto-drafted rosters your managers approve. Hotels cut 6–12% of labor cost without cutting guest service — and nothing writes back into Shiji.
Calculate your labor savings →Book a demo with your own Shiji data
Shiji Group · vendor website: shijigroup.com ↗
Shanghai, 320 rooms: the desk staffed to the curve Shiji already knew
Corporate arrivals cluster after 17:00; the flat roster paid seven agents from 08:00 and still queued the evening peak. Staffed to the hourly arrival forecast, the same total headcount moved to where guests actually arrive.
The data behind this chart
| Hour | Staff required | Old manual roster | Demand-matched roster |
|---|---|---|---|
| 08:00 | 4 | 7 | 4 |
| 10:00 | 3 | 7 | 3 |
| 12:00 | 5 | 7 | 5 |
| 14:00 | 5 | 7 | 5 |
| 16:00 | 7 | 7 | 7 |
| 17:00 | 10 | 7 | 10 |
| 18:00 | 11 | 7 | 11 |
| 19:00 | 10 | 7 | 10 |
| 20:00 | 8 | 7 | 8 |
| 22:00 | 5 | 7 | 5 |
Worked example, figures indicative, 320-room city property.
Put your own numbers on this → the calculator takes two minutes and needs no email — or see pricing first.
The workforce management layer on top of Shiji
Three modules on one read-only data feed, measured against your own baseline. Labor runs 32.4% of hotel revenue (CBRE, US benchmark); demand-matched scheduling recovers 6–12% of it.
AI demand forecasting
Arrivals, departures and covers forecast per department, per hour, from live Shiji pickup.
How forecasting works →AI staff scheduling
Draft rosters at the lowest cost that still hits your service standard; managers approve every schedule.
How scheduling works →PMS data insights
Revenue, labor cost and productivity per property on one live screen, straight from Shiji.
See the insights →Go live in three steps
Read-only API access · counts and dates only, no guest-identifying data · revocable at any time.
- Step 1
Approve read-only API access
Your Shiji administrator approves a read-only API connection scoped to the pilot properties, a short configuration task guided by our onboarding team, revocable at any time. Nothing writes back to Shiji.
- Step 2
Backfill and calibrate
HotelCadence pulls up to two years of Shiji reservation history and calibrates hourly demand forecasts per department against your own actuals, blended with external signals such as weather, day of week and public holidays like Golden Week and Chinese New Year.
- Step 3
Go live and measure
Forecasts refresh as pickup changes and draft rosters flag shifts that no longer match demand. Managers approve everything, and results are measured monthly against your agreed baseline.
Shiji integration questions, answered
How long does a Shiji integration take?
Your Shiji administrator approves read-only API access; most of the setup is backfilling reservation history and calibrating the forecast against your actuals. Integration is included in the 90-day pilot (from US$7,500).
Which Shiji systems does HotelCadence connect to?
Shiji Enterprise Platform, the API-first cloud PMS, is the standard connection. F&B covers can flow from Infrasys, Shiji’s cloud POS, where it is deployed. Properties on legacy Shiji installations are scoped individually in the integration audit before any commitment.
Can data stay in mainland China?
Yes. For portfolios with mainland China properties, regional hosting keeps operational data in-region, consistent with PIPL obligations. The connection is read-only either way, and no guest-identifying data is required for forecasting or scheduling.
Is HotelCadence a certified Shiji partner?
Certification listing is in progress. HotelCadence connects today through the documented Shiji Enterprise Platform APIs, the same mechanism Shiji’s own integration ecosystem uses, so nothing about the connection waits on the listing.
We run Shiji in China and OPERA Cloud elsewhere, does that work?
Yes, and it is the typical setup for chains in the region. Shiji properties connect via the Enterprise Platform API, OPERA Cloud properties via OHIP, Mews and Cloudbeds via their own APIs, and everything rolls up into one portfolio view with the same per-department, per-hour forecasts.
Why does scheduling matter more in a market where RevPAR is falling?
China closed 2025 with RevPAR down around 3%, so profit has to come from cost discipline rather than rate. Labor is the largest controllable cost, and demand-matched rosters make it flex with occupancy instead of staying fixed while revenue falls.
Calculate your labor savings →Book a demo with your own Shiji data
The calculator takes 2 minutes, no email needed · the demo is 30 minutes with read-only access · no obligation. Different PMS at other properties? See all integrations.