What a unified inbox for short-term rentals should actually do

Airbnb, Booking.com, WhatsApp, email and calls in one queue

6:05pm. Inside twenty minutes: an Airbnb guest asks about parking, a guest on WhatsApp asks what time check-in starts, and a corporate booker emails about next month. The first two are in the house manual, so the AI answers them. The email waits in the same queue for the bookings team in the morning. Priya hasn't been needed yet.

In Base360, email (Gmail, Microsoft 365 or your own domain), SMS, WhatsApp, booking-channel messages from Airbnb and Booking.com, and replies to Instagram and Messenger messages all land in one queue. So do phone calls, answered in the browser, with the recording and transcript kept in the guest's conversation, so nobody has to ask the guest to explain it all again. Comments on your Instagram, Facebook and TikTok posts have their own space next to it, so public replies don't get mixed up with private ones.

One thing to know: each channel has its own rules. Instagram and Messenger conversations can be answered but not started from the inbox, and booking-channel replies depend on the channel allowing a reply on that booking.

Separate inboxes for each team, city or portfolio

6:30pm. Lisbon is busy too, and so are the landlords. None of it reaches Priya tonight, because none of it belongs to the London shift.

In Base360 an administrator builds as many inboxes as you need from the channel accounts, users, properties and property groups each one should cover. A London inbox holds the London numbers and the London team; an Owners inbox holds landlord conversations. Notifications for a new message go to the person the conversation is assigned to first, then their department, then the rest of that inbox's team, so an on-call shift only hears about its own work. Everything still appears in All conversations for managers, and access is permission-based, so a team that covers one portfolio can be limited to that portfolio's messages.

Every message matched to a reservation, and honest about the rest

6:50pm. The first time Priya is needed. A WhatsApp from a number nobody recognises: "We're outside, the code isn't working." Tonight's check-in writing from a friend's phone, or someone who hasn't booked at all?

A good inbox doesn't guess, and neither should its AI. When a phone number or email clearly matches one reservation, Base360 proposes that booking. When it doesn't, the conversation stays Unmapped instead of being linked to the wrong guest. Priya checks, links it to the right stay before any door code goes out, and the guest is inside in minutes.

Sometimes it turns out to be a new enquiry from someone who hasn't booked yet. Those deserve a quick answer too, and they're easy to miss if the inbox only shows messages it can match.

The guest, the stay and the property beside every thread

7:15pm. Sam Carter, checking in to Demo Apartment 02 today, asks for two more nights at the end of the stay. The AI replies with the availability and the rate. Sam says yes, and the extension is a decision the team keeps, so it comes to Priya.

Priya doesn't need to ask anyone about Sam. The panel beside the conversation shows a returning guest, what they've spent overall and on this booking, payment still open and pre-check-in with ID not started. The guest portal link and the property's details, including check-in instructions and house manuals, are one click away, alongside previous bookings, open tasks and past calls. At the top, an AI summary of the conversation so far saves reading the whole thread to find out what's already been promised.

The queue helps with order, too. Each conversation card shows the stay stage, the property and the dates, and can be filtered by stay stage, channel, booking source or who it's assigned to, so a guest checking in today comes before one arriving in three weeks.

AI replies at the level you choose

Why did the parking question never reach Priya, while Sam's extension did? Because that's how the London inbox is set up.

In Base360 you choose how much the AI does for each inbox: no suggestions, Draft & Propose (the AI writes, a person checks and sends) or Auto-reply (eligible replies go out on their own). A sensible way in is to start with drafts and turn on auto-reply once you've seen the answers are right. You can pause the AI for an inbox at any time.

The AI answers from what you've given it: the recent conversation, your guest-facing knowledge, your house manuals, and live availability and rates. When it can't answer safely, it shows why and passes the conversation to your team rather than making something up. Analytics show the share of conversations it handled and the questions it couldn't answer, which are usually the house-manual pages you should write next. Every page you write is one more kind of message that stops reaching your on-call person.

The same goes for your team. Agents Performance shows response, messaging, call and resolution figures for the whole team or one person over any period, so a slow evening shows up in your own numbers before it shows up in your Airbnb response rate.

If you want to see where the AI should stop and your team should decide across a whole stay, we walk through it in AI-powered guest support and operation flow.

Guest messages that turn into maintenance and cleaning tasks

7:45pm. "The toilet isn't flushing properly." It's a guest message, and it's also a maintenance job. In most operations someone copies it into another tool, and some detail gets lost on the way.

Here the AI proposes the task right under the message, with the property, the room and how long the guest is staying. Priya gets one notification, checks the details and approves. Approving a maintenance proposal opens the task form already filled in; a cleaning proposal becomes a cleaning report. Nothing is created without a person saying yes, and the task shows up in the guest's panel as well as in your maintenance and cleaning views. If the AI doesn't propose one, Priya can create the maintenance or cleaning task by hand from the same panel.

Internal notes, owners and follow-ups nobody misses

8:10pm. Sam's extension is agreed, but the payment isn't through yet, and Friday's clean depends on it. Priya won't be on call on Friday. Someone else has to know.

An internal note stays inside your team and is never sent to the guest. Priya mentions a teammate, pins the note to the top of the thread for seven days, and assigns the conversation to John Doe for the morning. Every conversation has one owner, a person or a department, so it doesn't drop when someone is off, and several can be assigned or closed at once.

Waiting on a guest or a contractor? Snooze the conversation for an hour, four hours or a time you choose, and it comes back when it's needed. Once the issue is resolved, close the conversation; you can reopen it at any time.

WhatsApp's 24-hour rule, handled with templates

8:40pm. Tomorrow's checkout in another flat has a clean booked straight after, and the guest hasn't confirmed their time. That guest last wrote three days ago, and on WhatsApp that matters: free-form messages, photos and voice notes can only be sent within 24 hours of the guest's last message, a timer Meta calls the customer service window. After that, you need a template WhatsApp has approved.

Base360 shows when the window has closed and offers the approved templates: a checkout reminder, a payment link, a message offering an extension. Priya picks the checkout reminder, fills in the guest's first name, checks the preview and sends. WhatsApp, SMS and email templates live in one template library, and a new email or WhatsApp conversation can start from the inbox instead of another app.

Replies in the guest's language

9:05pm. A guest arriving next week writes in Spanish: can they drop their bags at 9am, before check-in? That depends on the cleaning schedule, so the AI drafts a reply and leaves it for a person. Priya doesn't speak Spanish, and doesn't need to.

In Base360 you can translate a single message, the whole thread, your own draft or the AI's proposed reply, check it and send, without copying anything into a translation tool and back. Translation currently covers English, French, Spanish, Arabic, German, Italian, Simplified Chinese and Portuguese.

Call logs you can review, and calls that reach the right people

9:30pm. The last call of the night: a guest unhappy about noise from the street. A call like this should reach a person, and it does. The London call group rings for the London shift, Priya takes it in the browser and offers a quieter room if one is free tomorrow.

The next morning, John Doe doesn't need to ask how the evening went. Every call is kept in Call Logs, with a Missed view for the ones nobody answered. Each record shows who took the call, how long it lasted, the numbers involved and the sentiment, and, where they exist, the recording, the transcript and an AI summary.

Call groups decide who rings for incoming calls: which people, which numbers and how the call is routed. Ours rings everyone on the shift at once, and if nobody answers it goes on to a manager, so a missed guest call never goes unnoticed. Team members can be limited to the numbers of their own group.