Markus Breth

Product · AI agent under approval

The agent proposes. A human approves.

Rates, bookings, guests, cleaning and billing in one place. The agent watches the whole operation, proposes changes and lays out what each proposal rests on. The property decides per kind of action what may run without asking — rates never reach the booking channels without approval. In production since July 2026 at a hotel and rental group in Austria, on their own server, with a local language model.

  • In production since July 2026
  • Over 700 tests
  • Your own server
  • DE / EN

The problem

Running a property is an endless run of small decisions.

None of them is hard. All of them together are too many — and the ones left undone cost the most.

  • 01Rates go stale by the hour. Miss a short-notice adjustment and you sell too cheap, or not at all. By hand, nobody keeps up across every unit and every day.
  • 02The small stuff lapses in silence. The deadline for a damage claim, the missing guest registration, the invoice after check-out: no reminder arrives on its own — until it is too late.
  • 03AI you cannot watch. Tools show *that* something was changed automatically. Why, based on what and whether it worked stays open — and therefore unverifiable.

How the agent works

Observe · propose · approve · measure.

Every one of these steps leaves a row in the log. The dashboard is nothing but the view onto it.

01

Observe

Bookings, rates, reviews and dates are pulled in from the booking channels and evaluated continuously.

02

Propose

Every proposal comes with before/after, its reasoning and the evidence it stands on — not as a bare instruction.

03

Approve

A human sees before and after and decides with one click. Without approval no rate change reaches a booking channel.

04

Measure

After a rate change the system checks days later whether the gap got booked — and writes the result back.

What the agent proposes

One head for the whole operation.

Every kind of action is governed on its own: execute directly, submit for approval, or never propose at all.

◆ RATES

Rates

Spots empty nights and proposes an adjustment. Once approved, the rate goes through the channel manager to Airbnb, Booking.com and other channels.

◆ GUEST

Messages & extras

Drafts for guest messages and fitting offers such as late check-out or early check-in. What goes out is what was approved.

◆ REVIEWS

Reviews

Sorts new reviews by topic and severity and drafts a response to critical ones in the property’s voice — submitted, not published.

◆ TASKS

Cleaning & maintenance

Creates cleanings to match departures and generates recurring maintenance and inspections from templates — instead of a second list.

◆ INVOICE

Invoices

After check-out, the invoice with Austrian VAT and city levy — without charging twice when the channel already collected the levy.

◆ ALERTS

Metrics & deadlines

Flags when forward bookings or cancellations drift from target, and reminds you before the deadline for an Airbnb damage claim runs out.

Guests & team

Fewer questions, less paper.

Three surfaces from one system: the cockpit for the office, a page for the guest and an app for the staff.

  • →Online check-in The guest fills in the registration form on their phone before arrival, including fellow travellers and signature. The system sends the reminder; the property prints it as a PDF when needed.
  • →Chat & enquiries Messages from the website chat land in one inbox. At the push of a button a suggested reply is ready, in the guest’s language.
  • →Self-service invoice Via a personal link or with booking number and surname on the website — the invoice arrives as a PDF.
  • →Staff app Cleaners see their jobs on their phone, tick off checklists, upload photos and report damage straight to the office. Access can be revoked at any time.
  • →Morning briefing & agent chat A short message every morning with arrivals, departures, cleanings and pending approvals. Questions about the operation in plain language — proposals from the chat go through approval too.

Admin & data

Billing comes along. The data stays in-house.

AI Hotel OS runs on your own server. All data and everything learned stays in your own database.

  • ✓Owner statements Per owner and month: revenue, fees, commission, cleaning and payout, as a PDF.
  • ✓Export for the accountant Monthly report including city levy, as CSV and in BMD format, chart of accounts configurable per business.
  • ✓GDPR Data access requests as an export per guest. Anonymise instead of delete, so invoices still meet the retention duty.
  • ✓Access & audit log Roles with graded permissions, two-factor sign-in and a log of every change. Daily backups.
  • ✓Integrations Hospitable as channel manager, Microsoft 365 for sending mail, Telegram for alerts. An API and webhooks for your own systems.

Where it stands

In operation, not in a brochure.

700+
automated tests
8
kinds of action, each governed on its own
0
rate changes sent to booking channels without human approval
3
surfaces: cockpit, guest page, staff app

Best seen live.

AI Hotel OS runs on your own infrastructure and is tailored to your property, channels and processes. In a live demo I show how a proposal comes about, what approval looks like and what happens next.