Integrations

One warehouse. Every tool you already run.

We read your systems into a warehouse that belongs to you. Each tool is connected by the role it plays, so the PMS is the PMS whether it is Guesty or Hostaway, and swapping one for the other later is a connector, not a rebuild. Read only, every day, with the account verified before a single row moves.

Read only, by constructionYour own warehouseOne connector per toolSet up in less than a week

How it is organised

Tools are connected by role, not by name.

Every report and detector reads from the role. That is what makes the system portable across stacks, and what keeps a change of tool from turning into a rebuild.

RoleWhat we read from itLive todayNext
Property management systemListings and the daily roster, reservations of every status, folio lines, nightly fares, payments, the money modelGuestyHostaway
Pricing and revenue managementRecommended rates, market context, minimum stay by date, daily availability snapshotsWheelhousePriceLabs
CRMDeals and full stage history, contacts, owners, tasks, email metadata. Never email bodiesHubSpot
Guest messaging and operations AIConversation metadata, escalations, response times, task events. Never message contentConduit
Field operationsCleaning, inspection and maintenance tasks, durations, issues raisedBreezeway
Ancillary revenue and experiencesOrders per stay, attach rate, commission, what sells whereEpicurate
Email marketingCampaigns, audiences, send and engagement metadataMailchimp
Guest data captureCapture volume by property, consent state, downstream campaign metadataStayFi
AccountingOwner statements, receivables, the cash side of every bookingQuickBooks, Xero

Every tool

Where each one stands.

A tool gets its own page when its connector has shipped and there is something real to say about it. Until then it is listed here with its role and its status, and nothing more, because a page with nothing behind it helps nobody.

Next

Next

Hostaway

Property management system

Same role as Guesty, same staging contract downstream. One connector, and every report and detector works unchanged.

Next

PriceLabs

Pricing and revenue management

Same role as Wheelhouse. Recommended rates, market data and minimum-stay rules by date.

Planned

Planned

Conduit

Guest messaging and operations AI

Conversation and escalation metadata, response times, and the operational tasks its agents raise. Never the message content.

Planned

Breezeway

Field operations

Tasks, durations, issues. One caution from experience: field-ops tools rarely carry labour cost, so measure fill before building a metric on it.

Planned

Epicurate

Ancillary revenue and experiences

What sells per stay, attach rate by property and season, commission earned. The revenue between a reservation and checkout.

Planned

Mailchimp

Email marketing

Campaigns, audiences and engagement metadata, so a repeat-guest offer can be measured from send to booking.

Planned

StayFi

Guest data capture

Capture volume and consent by property, and what the downstream campaigns did. Contact details stay where the AI cannot reach them.

Planned

QuickBooks, Xero

Accounting

The cash side of every booking, so recovered revenue can be confirmed as collected, not just booked.

What a connector is

Rules every connector follows, whichever tool it reads.

Read onlyBy construction, not by promise. Most vendor keys can write. Ours cannot. The session refuses every write verb before it leaves the machine.
Verify the accountA valid key on the wrong account returns a normal 200. Every connector checks it is looking at your account before it loads a single row.
Never replaceRows merge on a natural key and carry the date they were last seen. A record the source has forgotten keeps its history; a line deleted at the source stops counting.
Snapshot dailyRosters, availability and rates are photographed every day in your local timezone, which is what makes same-time-last-year possible a year later.
Money is declaredEvery money column is typed at the point it is read. A loader that guesses the type from the first rows will corrupt the decimals that come later, silently.
Fail loudlyIf the rows collected are fewer than the source says exist, the load fails. A partial revenue load that succeeds quietly is the worst outcome there is.
Poll, then webhookEvery source gets a scheduled reconciling pull. Webhooks only improve latency; they never replace the poll.
MinimiseMetadata, not content. Sensitive identifiers are refused by name before the request is made, and asserted absent afterwards.

Run something that is not on this list?

Tell us what it is and what role it plays. If it has an API, it is a connector and a staging model, not a project. If it does not, we will say so.