Managed guest messaging — hotels & rentals

Guest messaging support, managed by Relaytask

AI answers repetitive guest questions using property knowledge you approve. Relaytask specialists handle the conversations that need judgment, within the coverage window your plan buys. Anything needing your authorization or someone on the property follows the escalation rules you agreed.

A Relaytask support specialist handling a guest conversation at a workstation

Most of the guest inbox is the same few questions

They arrive across whichever channels your guests already use, at whatever hour suits the guest, and they do not stop because the desk has gone home.

Arrival and access

Check-in times, directions, door codes, parking, late arrivals and the message that starts “I'm outside and…”.

Property questions

Wi-Fi, appliances, amenities, house rules, pet policy, early check-in and late check-out requests.

Booking questions

What is included, what the cancellation policy says, and what a guest has already been told — answered from what you have approved.

The same question, again

The repetitive share of the inbox that arrives at every hour and rarely needs a person to read it first.

AI when it can. A person when it matters. You when it is your call.

The examples below are illustrative, not customer cases. What decides the route is the rule set you approve at setup — not a guess made in the moment.

AI answers

Approved knowledge covers it

“What time can we check in, and where do we park?”

Answered from the property knowledge you approved, in the tone you set, on the channels configured in your setup.

A specialist takes over

Judgment, empathy or an exception is involved

“The code isn't working and we've been standing here twenty minutes.”

A trained RelayTask specialist picks up the conversation with its history and context attached, during the human coverage window on your plan.

It comes to you

Your authorization or someone on site is needed

“We'd like a refund for the first night.”

Refunds, compensation, reservation changes and anything physical on the property route to your named contact under the escalation rules agreed at setup.

What has to be agreed before a single guest is answered

An AI that answers from nothing is a liability. Everything it is allowed to say comes from knowledge you approved, and everything it is not allowed to decide has a route to someone who can.

Step 01

Approved knowledge

Your property information, policies and standard answers — written down, reviewed by you, and the only thing the AI may answer from.

Step 02

Permitted actions

What may be answered outright, what must be escalated, and what nobody may commit to without you.

Step 03

Escalation contacts

Who is called, for what, and on which route — including the emergency path that never relies on a queue.

Step 04

Coverage window

The hours your plan buys human availability for, chosen at onboarding.

Step 05

Ongoing upkeep

Knowledge, policies and routing maintained as your properties and rules change. That upkeep is part of the managed service, not your task list.

What coverage actually means here

AI responds around the clock, but only within the knowledge you approved. Human availability is the window your plan buys — outside it, AI keeps responding where it appropriately can and queues anything needing a person for the next window. It never implies somebody is standing by when nobody is.

Continuous live human availability exists, but it is an Enterprise arrangement scoped around your operation rather than something every plan quietly includes.

Coverage windows by plan

Not included, and deliberately so

  • Refunds, credits and compensation decisions
  • Reservation changes, unless you have explicitly authorized them and the system supports it
  • Anything physical on the property
  • Maintenance execution
  • Emergency response itself — emergencies follow your routing, not an AI queue

These sit with your authorized people because they are decisions about your property, your money and your guests — not because the operation cannot see them coming. Each one has a documented route.

Choosing between the two

A managed inbox, or your own trained people

Both cover guest conversations. The question is how much of the operation you want to keep, and how much of the work sits outside the inbox.

This page

Hand the guest-message operation over

Choose this when the load is messaging volume rather than broad operational work, when a large share of it repeats, and when you would rather approve the rules than run the process. Relaytask manages the AI, the specialists, the routing and the upkeep.

How Managed Support works

Dedicated team

Put your own trained people on it

Choose this when the work runs well beyond messaging — reservations administration, vendor and turnover coordination, reviews, back-office records — and you want named, office-based people inside your own process, supervised on the floor.

Working out which conversations should stay automated? What AI can and cannot handle in a guest inbox sets out the split conversation by conversation.

Running properties on behalf of owners rather than your own? Support for property management companies covers the owner side of that work.

FAQ

Common questions

The channels that are technically supported and connected during your setup. Managed Support covers seven text channels — live chat, WhatsApp, Facebook, Instagram, email, SMS and Telegram — and which of them are live for you depends on what your operation actually uses and what can be connected. Voice is not part of the seven-channel Managed Support offer.

See the full Managed Support operation

Only if your plan buys coverage at that hour. AI keeps responding around the clock within approved knowledge, collects details, and queues anything needing a person for the next human coverage window. Continuous live human availability is an Enterprise arrangement. Safety, access and other emergency-type issues follow the emergency routing agreed at setup rather than waiting in a queue.

No, not on its own. Reservation changes need both your explicit authorization and a system that supports the change; refunds, credits and compensation stay with your authorized people. Those requests are recognized, gathered and routed rather than acted on.

A chatbot product gives you software and leaves you to staff what it cannot answer. Here the software, the AI setup and knowledge configuration, the trained people who take over, the routing, the quality oversight and the ongoing knowledge upkeep are all part of one managed service. The practical difference is who does the work when the automation stops being enough.

A dedicated assistant is a person inside the operation you run — you set the process, the hours and the supervision, and the whole inbox is read by a human. This is a managed operation where AI takes the repetitive share first and RelayTask runs the support process around it. Operators who want their own trained people, working across reservations and back-office work as well as messaging, are usually better served by dedicated hospitality support.

Compare dedicated hospitality support

Managed Support plans are matched primarily to your monthly guest-conversation volume. Portfolio size, human-escalation usage, coverage window and operational complexity are confirmed during setup. Published monthly plans and what each includes are on the Managed Support page.

Managed Support plans

Only what the agreed work requires, under access you grant and can revoke. RelayTask is an independent outsourcing provider and is not affiliated with or endorsed by any booking platform.

Get Started Today

Tell us what your guest inbox looks like

Roughly how many guest conversations a month, which channels they arrive on, and the hours you need a person available. That is enough to size the plan.

  • Free support audit
  • AI available 24/7 within approved knowledge
  • Human coverage sized to your plan
  • No separate chatbot software subscription
Two colleagues celebrating at a desk in a bright office

5-day onboarding setup

Trained, supervised, office-based