Every email to a boutique hotel's reservations inbox is classified by AI. Booking requests get a reply draft with live availability and rates from the property management system, vouchers and invoices are routed, and same-day arrivals are flagged at the top.
Typical results
≈3 hrs/day
Team time freed across two reservations agents
Under 1 hr
First reply to new booking requests
6
Email types sorted automatically
Figures reflect typical results for this kind of automation.
The challenge
Two reservations agents handle about 60 emails a day: booking requests, date changes, cancellations, group inquiries, travel agent vouchers and supplier invoices, all in one inbox. Each booking request means opening the property management system to check availability and rates before typing a reply. Urgent messages about same-day arrivals can sit under a pile of routine mail.
Before and after
Minutes of handling per email
−50%Hours to first reply on booking requests
−80%Same-day arrival emails missed per month
−100%The automation
AI reads each incoming reservations email and labels it as a new booking request, change or cancellation, group inquiry, travel agent voucher, invoice or other. For booking requests, it checks availability and rates in the property management system and drafts a reply in Gmail or Outlook for the team to review and send. Vouchers are logged against the matching booking, invoices are forwarded to accounting, and same-day arrivals are flagged at the top of the queue.
How it works
- 01
Email arrives
Every message to the reservations inbox is read as soon as it lands, including attachments.
- 02
AI classifies it
It is labeled as a booking request, change or cancellation, group inquiry, voucher, invoice or other.
- 03
Availability checked
For booking requests, open rooms and rates for the dates are pulled from the property management system.
- 04
Draft or routing
A reply draft is prepared for the team, vouchers are logged on the booking and invoices go to accounting.
- 05
Sorted queue
Agents work from a queue sorted by type, with same-day arrivals flagged at the top.
Starts from
- Reservations inbox
- PMS availability and rates
Delivers
- Reply drafts for review
- Vouchers logged on bookings
- Invoices sent to accounting
The impact
The team works from a sorted queue with drafts ready, and handling time per email falls by about half, freeing roughly 3 hours a day between the two agents. Booking requests get a first reply in under an hour instead of around five, and same-day arrival messages are no longer missed.
Questions about this automation
Does the AI send replies to guests on its own?
No. It writes a draft in the team's mailbox with the dates, room types and rates it found, and an agent reviews, edits and sends it. The hotel can later let simple categories, such as a cancellation acknowledgment, go out automatically, but every booking reply stays with a person.
How does it check availability and rates?
It reads the requested dates, number of guests and room preferences from the email, then queries the property management system through its API for open rooms and the current rate plan. If the dates are unclear or the hotel is full, the draft says so and suggests alternative dates for the agent to offer.
What happens with emails it can't classify?
Anything that doesn't fit a category with confidence is labeled other and placed at the top of the queue for an agent, with a one-line summary. Mail in other languages is classified the same way, and the draft is written in the guest's language. This is a multi-step build, usually rolled out in stages.
