

Pilot gate (only you see this)
Or three quick facts here. Skip any you like; you can change everything later in Setup.
How do you want to run your team?
The team is paused. Nothing runs and nothing finishes on its own; new requests wait.
Tell the team what you need. It picks the right bot, sets the priority and asks if something is missing.
Photos, PDFs, spreadsheets, CSV or text. The bots read them with the request.

Your own line to the operator, a real person with hospitality experience. Only for what a person must decide (money, legal, safety, an upset guest, the unusual case). You see progress and a clear "done when".
Every request, newest first, with who approved what.
Take your work with you
What the team did and what it moved, from the work in this office. Numbers are counted, not guessed.
The more you tell the team, the better the work. With nothing here, the bots ask you or leave a [blank] for you to fill; they never guess a fact for a guest.
Fast start: fill from your listing
Paste the text of your Booking.com or Airbnb listing, your website, or your house rules, or upload them (PDF, text, screenshots). The team fills the empty fields below with what the listing says and nothing else. Check them, then press Save property.
This changes how the bots talk to you, never what guests read.
Price per night by season. The team quotes the normal price and never invents one. Lowest is optional: how far you would go down when you allow a discount; the team never goes below it. Leave the unit empty if all units cost the same.
A guest on the phone? Two dates, and the app works out the total from your saved prices. No AI, nothing is sent.
It hangs on your office wall, and the brand colour is taken from it. You can still change the colour.
Design my office from my property
Add 1 to 3 photos (terrace, a room, the view) or describe the place. Claude picks the colours, the view in the window and the details. The bots and desks stay the same.
The clock, the window and your bots' sense of "today" and "tonight" follow this time zone: sunrise, day, sunset, and at night the moon, the stars and the office lights.
The facts guests ask about most. Anything left empty, the team asks you about instead of guessing.
It shows on your Today card and every bot keeps it in mind. Change it any time.
On its own, the team still stops for you on refunds, compensation, damage, legal matters, upset guests and anything the operator must see. The bots never send anything to guests themselves: finished work is ready for you to use.
Suggestions show on your Today card. Nothing starts until you tap Yes, so they never cost anything on their own.
Off unless you tick it. If ticked, the changes you make to drafts and your ratings may be used, with names, contact details and booking numbers removed, to make the bots better for every property. Untick any time; from then on nothing new is used. Without it, what your bots learn stays in your office.
Upload your latest numbers about once a week: a booking export from your system (CSV or Excel), a rate report, a profit and loss or cost list, or a screenshot of your competitors' prices. Rev, Sal and Fin read them in every answer. A new upload replaces the old one.
Your calendar
Booking.com, Airbnb and most booking systems give a free calendar link (export or iCal). Paste it here and the team sees today's arrivals, how full you are and your empty nights. Guest names are not in it.
Your rules
Rules the team follows in every answer, learned from your ratings. Remove any that no longer apply.
Decision log
When you approve an answer from Rev, Sal or Fin, the decision is saved here in one line. They read their last 10 decisions before every answer, so they remember what was decided and why. Remove any that no longer apply.
Hire the roles you need, as many as you need, and release a hire when you don't. Each hire gets a desk in the office.
How you work with the team
Only you see this. Playbooks are the expertise each bot follows. Approved drafts that were edited become examples the bots learn from.
Expert requests
Pilot scoreboard
How useful the owner found each finished draft. This is the number the pilot is judged on.
Lessons from the owners' fixes
How often each bot's drafts were ready to use, week by week. Once a week, the drafts the owner had to fix become one proposed change per bot; nothing changes in a playbook until you add it.
Lessons across all offices
Only offices that ticked 'Help improve the team', and only drafts fixed after the tick. Names, contact details and booking numbers are left out. Accepted lessons go into the standard playbook of every office with the next build.
Test the team
Real requests with written pass points. The team answers each one and an independent check grades it. Core 12 is the pilot gate: 10 of 12 passed with no invented facts. The other sets test one bot at a time.
Paste the guests' messages only (not your answers), with a line of three dashes (---) between them. Email addresses and phone numbers are blanked before saving; replace names yourself. Each message becomes a case in 'Your cases', graded on: answers what was asked, nothing invented, prices only from your table, never confirms free dates, guest's language, sendable after light edits.