Run the venue. Own the story.
Finance the asset.

HotelStack.ai isn’t a bundle of apps that share a login. It’s one data spine and one intelligence engine, read through different lenses — so the shift report, the market view and the board pack can never disagree with each other.

How it fits together

Your hotel’s paperwork comes in once. It becomes one set of numbers. Every product, and the AI, reads from that set.

1 · Your paperwork comes in2 · It becomes one set of numbers3 · Everyone reads from that one setNight auditEmailLiveTakingsExportExport-readyRosterCSV · SFTPBuiltPurchasesRESTConnectedSalesExport · APIAssessedAnything elseMCPLiveread oncepersistedMetric layerThe Exemplar · 214 rooms · modelledRooms revenue1,042,880Occupancy178.4%ADR$268F&B revenue812,300Rooms payroll25.0%Cost of sales31.2%EBITDA · TTM2$4.20mserved with provenance · computed onceViPthe duty manager · tonightHiPthe development lead · the marketAiPthe asset manager · the ownerPiPthe owner and board · the estatecites the row it readBoardroomthe AI reads the same numbers · cites the row12 The harnessparsers · metrics · citations · audit log · any model inside1 · Your paperwork comes inNight auditEmail · LiveTakingsExport · Export-readyRosterCSV · SFTP · BuiltPurchasesREST · ConnectedSalesExport · API · AssessedAnything elseMCP · Liveread once · persisted2 · One set of numbersThe Exemplar · 214 rooms · modelledRooms revenue1,042,880Occupancy178.4%ADR$268F&B revenue812,300Rooms payroll25.0%Cost of sales31.2%EBITDA · TTM2$4.20mserved with provenance · computed onceViPduty managerHiPdev leadAiPasset managerPiPowner + boardBoardroomthe AI reads the same numbers · cites the row12The harness · parsers, metrics, citations, audit log3 · Everyone reads from it
Modelled data · The Exemplar · 214 rooms · not a live hotel · statuses as the integrations table states them
Operate

ViP

Run every shift with the whole venue in view — POS, bookings, tasks, kitchen, signage, sensors.

Explore →
Invest

HiP + AiP

The market lens and the capital lens — greenfield analysis, benchmarking, covenants, valuation.

Explore →
Own

PiP

The whole estate for owners, boards and group finance — tree, league tables, exceptions first, Boardroom’s portfolio read. In preview.

Explore →
Present

Marquee + Signage Studio

Marquee builds venue sites on your domain, SEO-linted before publish. Signage Studio puts your Canva pages on the venue’s screens with a rule engine and snapshots.

Live
Sell

Events

Selling the space is the next pillar — in design now, shipped when it can run our own events.

In design

Independent by construction.

Vendor independence is not a feature we added. It is how the platform is put together. The constants are the harness and the governed metric layer. Everything on either side of them, the model that answers, the systems that feed it, the cloud it runs in, is replaceable without touching your numbers or your history.

The test we hold ourselves to: could you change the model, swap a PMS, or move clouds, and have every figure, every citation and every audit entry survive it? Yes, by design, and the table below says which changes are routine today and which are by arrangement.

The AI

The model is a per-task setting behind the harness, today frontier models from Anthropic. Nothing in your data, the metric layer or the audit log depends on which model is inside. Change it, or bring your own.

How the harness treats the model →

The systems you run

We read what your PMS, POS, payroll, procurement and finance systems already produce. Replace one and the metric layer, its history and the board pack carry on; only the parser changes.

Every system, channel and status →

The infrastructure

Containers and Terraform, hosted in Sydney today, cloud-agnostic by decision. Your cloud, your data platform or your premises, by arrangement, with the same rules in force.

For enterprise estates →

Keep the systems you like. We read what they already produce.

HotelStack.ai does not ask you to replace anything. Whatever your PMS, POS, reservations, events, workforce, procurement or finance platform is, we take its output the way it already leaves the building: an emailed report, a file drop, an API, or an MCP tool your own assistant can call. The channel column is the point. The status column is honest about which systems we have already wired and which we would wire on day one.

A statement that arrives as a PDF is parsed to the same governed metrics as one that arrives over an API. Provenance records which.

If we built it, there is no per-seat licence. Every person in the venue can use our own apps, from the chef’s tablet to the owner’s board pack. You never pay for a login.

We built them because the tools a venue actually runs on, tasks, prep lists, food safety records, training, signage, the website, arrived as a stack of separate subscriptions. Each was priced per seat or per screen, so the cost climbed with every person who used it, and each held a slice of the day that never reached the numbers. So we replaced them, one at a time, in our own venues first. Our chefs print prep labels from the kitchen tablet. Our food safety records are kept here, to council-inspection standard, instead of in a separate compliance app. Our own venue websites run on Marquee. When something is clumsy, the people who tell us are on shift in our own kitchens, restaurants, front desk and housekeeping.

They are hospitality apps, not office software with a hospitality skin, and they run across the whole building. In the kitchen, a prep list understands use-by dates and allergens, and the food safety records are kept to council-inspection standard. In the restaurant, the checklists follow opens and closes and the screens on the wall run from Signage Studio. Front of house keeps its training and documents in the same place, and housekeeping its room checklists. And because nobody pays for a login, everybody uses them, so the operating picture is complete rather than a sample of whoever had a seat.

Hospitality systems by category, the channel each uses to reach HotelStack.ai, and its status.
SystemChannelHow the data reaches usStatusLive: running for a venue today. Export-ready: any report it emails or exports is parsed to governed metrics today; no bespoke connector yet.
Built by us · included · no per-seat licensing
Tasks & checklistsNative · in ViPLive
Kitchen prep & labelsNative · chef tablet · label printersLive
Food safety recordsNative · replaces DidgeLive
Training & documentsNative · in ViPLive
Asset registerNative · in ViPLive
Maintenance logNative · in ViPLive
Stock & reconciliationNative · in ViPLive
Signage StudioNative · your screensLive
Marquee venue websitesNative · your domainLive
Marketing StudioNativeLive beta
Sensors & ambienceNative · our hardwareIn progress
Cold-chain monitoringNative · sensorsSpecified
EventsNativeIn design
Property management
Oracle Opera Cloud (opens in a new tab)Email · night-audit PDFLive
Mews (opens in a new tab)Export · API availableExport-ready
RMS Cloud (opens in a new tab)Export · API availableExport-ready
Protel (opens in a new tab)ExportExport-ready
Cloudbeds (opens in a new tab)Export · API availableExport-ready
Guestline (opens in a new tab)ExportExport-ready
Point of sale
Bepoz (opens in a new tab)Email · daily exportLive
Oracle Simphony (opens in a new tab)Export · API availableAssessed
Lightspeed (opens in a new tab)Export · API availableExport-ready
H&L (opens in a new tab)ExportExport-ready
Impos (opens in a new tab)ExportExport-ready
Square (opens in a new tab)Export · API availableExport-ready
Reservations
SevenRooms (opens in a new tab)APIScaffolded
OpenTable (opens in a new tab)Export · API availableExport-ready
ResDiary (opens in a new tab)Export · API availableExport-ready
Now Book It (opens in a new tab)ExportExport-ready
Quandoo (opens in a new tab)ExportExport-ready
Events & function sales
iVvy (opens in a new tab)APILive
Tripleseat (opens in a new tab)Export · API availableExport-ready
Event Temple (opens in a new tab)Export · API availableExport-ready
Momentus (Ungerboeck) (opens in a new tab)ExportExport-ready
Workforce & payroll
Tanda (opens in a new tab)CSV · SFTP · daily payrollLive
Deputy (opens in a new tab)Export · API availableExport-ready
Humanforce (opens in a new tab)ExportExport-ready
Employment Hero (opens in a new tab)Export · API availableExport-ready
Procurement
PurchasePlus (opens in a new tab)REST APIConnected · read-only
Ordermentum (opens in a new tab)ExportExport-ready
FoodByUs (opens in a new tab)ExportExport-ready
Finance & reporting
Calumo (opens in a new tab)File · month-end exportLive
Xero (opens in a new tab)Export · API availableExport-ready
MYOB (opens in a new tab)ExportExport-ready
NetSuite (opens in a new tab)Export · API availableExport-ready
Market & reputation
STR / CoStar (opens in a new tab)Report · market shareLive
TrustYou (opens in a new tab)Email · inbound feedLive
Shiji ReviewPro (opens in a new tab)ExportExport-ready
Google Business Profile (opens in a new tab)Export · API availableExport-ready
Lighthouse (opens in a new tab)Export · morning rate-shop reportExport-ready
HotStats (opens in a new tab)Report · comp-set P&L benchmarkLive
Visual analytics & AI
meldCX Viana ↓API · webhook · footfall as aggregates, never identityLive
Your AI assistant
Claude, or any MCP clientMCP · six read-only tools · PiP connectorLive
Anything elseEmail, CSV, SFTP, REST or MCP. If it leaves your system, we can read it.

Partner · vision analytics

Vision analytics with meldCX Viana. What people do in a space, never who they are.

meldCX is a Melbourne vision-AI company and our closest partner. Our venues are where new Viana features are tested, inside HotelStack.ai, so we see them early and help shape them. James leads IT and security at meldCX while building HotelStack.ai, so the integration is designed from both sides. For many estates it is the reason they call.

Viana does no facial recognition, blurs faces during processing and is trained on anonymised synthetic data. HotelStack.ai then takes aggregates only. The rule is enforced in code, three times, not in a policy document.

What you learn

Traffic, dwell, glances and impressions per zone per hour, from the cameras already on your ceiling or the signage screens already on your walls. Queue events land in the operating picture as they happen.

What never reaches us

No face, no identity, no per-person record. The per-observation identifier is stripped at the fetch boundary, refused at the database boundary and has no column to land in. There is deliberately no unique-visitor count, because deriving one needs the identifier we refuse.

Demographics, with a floor

Age group, gender and mood exist only as a breakdown of at least five observations in an hour. A thinner hour suppresses the whole breakdown, not just the small cells, so nothing leaks by subtraction.

Reported as aggregates on the Trust page, alongside the rest of how we treat guest-adjacent data. Viana at meldcx.com →

For enterprise estates

We bring the harness. You bring what you already have.

Some estates arrive with a data platform of their own, a model they already trust, or a rule that nothing leaves their cloud. The harness is designed to sit on top of what you have rather than beside it: the parsers, the governed metric layer, the audit log and Boardroom’s accountability layers are ours; the layer beneath them can be yours.

Whatever you bring, the rules hold: the model sees a minimum, role-scoped slice; every claim cites its row; confidence is capped by the evidence; writes wait for a person; every call is logged. Those are the harness. They are not options.

Your data platform

Snowflake, Fabric, Databricks, a lake or warehouse of your own. Tell us where your estate’s truth already lives; we would land the governed metric layer there, or read from it, so your analysts keep working where they are.

Your model, or your agent

Your own key, endpoint or region for inference, or your own agent reading the estate through the connector. The harness still mediates every call, so the honesty travels with it.

Your cloud, or your premises

A dedicated deployment in your own account, in an Australian region, from the same containers we run ourselves. Or the edge on your hardware, as the sensor and signage agents already are.

None of this is a checkbox on a plan. It is a conversation about your model, your host and your data, and where the line between ours and yours should sit. We would rather have it early.

Talk to us about your estate

Four lenses, one set of numbers.

Start with the lens that matches your seat at the table.