Review Response Agent
Replies that read like a person, and never argue with a guest
- 9hard rules
- Things it must never do, enforced in every run
- 5handover rules
- Conditions where it stops and gives the work to a person
- 5output fields
- A structured result, not a paragraph you have to re-read
- ≤4steps per run
- Bounded work: at most 3 tool calls, enforced by the runtime
Read-only by design. This agent has no path to sending, spending, publishing or committing anything. Where that limit is the product, removing it would remove the reason to trust the output.
Overview
Replying to reviews is high-volume, public, and usually done badly — because it is done angry, late at night, by the person the review is about.
This agent drafts every reply: in the guest's own language, in your voice, answering the specific thing they raised rather than thanking them for the feedback. Where a review names a real operational problem it says so in your report, because the same complaint appearing eleven times is worth more than any single reply.
What it will not do is the part that causes damage. It does not dispute a guest's account of their stay, because a public argument with a customer costs more bookings than the review did. It does not apologise for something you did not do. It offers no refund, upgrade or discount — that is your money and your decision. And it posts nothing: every reply waits for you.
Benefits
Every review answered
Including the ones that sit unanswered for weeks.
Nothing you would retract
No apology for something you did not do, no offer you did not make.
The complaints become data
You see what is actually going wrong, not just what to reply to.
What it does
Answers the actual complaint
Not "thank you for your feedback" — the specific thing they raised, addressed.
Never argues
It does not dispute a guest's account of their own stay. A public argument costs more bookings than the review did.
Offers nothing
No refunds, upgrades or discounts. That is your money and your decision, not a draft's.
Replies in their language
In the language the review was written in, and in your voice if you supply past replies.
Finds the pattern behind the reviews
The same complaint eleven times is an operational finding worth more than eleven replies.
Posts nothing
Every reply is a draft. What goes out under your name is approved by you.
What this changes
What this changes about your week
- It takes over
- Replying to every review, in the reviewer's language, without sounding defensive.
- Done today by
- The owner, late at night, about their own business.
- How often
- Continuously, and it is public.
- Typical manual effort
- A considered reply is ten minutes; most properties have weeks of unanswered ones.
Why Hospitality in particular. It never disputes a guest's account and never offers compensation. A public argument with a customer costs more bookings than the review did.
Runs unattended
Started by you or by an event, and it finishes on its own. Nothing waits for someone to be at a desk.
The same standard every time
The two-hundredth item is held to the bar the first one was. Consistency is the part people cannot sustain.
It cannot act on its own
Review Response Agent has no path to sending, spending or committing. That limit is why its output is safe to act on.
How it runs
This agent runs server-side through the PROMIVO runtime. Each run is logged step by step and every tool call is permission-checked before it executes.
- 1InputValidated against the agent input schema
- 2Modelclaude-sonnet-5 with automatic fallback
- 3Toolsextract_structured_data
- 4PermissionsRuns READ actions automatically; anything higher waits for approval
- 5OutputStructured result matching the output schema
- 6LoggingTokens, duration and every step recorded
Demo
Demo dataIllustrative sample output, abridged.
Input
{
"reviews": [
{
"id": "r-91",
"text": "…",
"rating": 2
}
],
"businessName": "Marina Suites"
}Output
{
"replies": [
{
"id": "r-91",
"draft": "Thank you for telling us about the noise from the works next door — you were right to expect a quiet room and we did not warn you. We have changed how we allocate rooms on that side while the work continues. If you come back, please contact me directly and I will make sure you are well away from it.",
"status": "drafted",
"language": "en",
"addresses": "Construction noise, and the absence of any warning at check-in."
}
],
"escalate": false,
"patterns": [
{
"count": 7,
"issue": "Construction noise on the north side",
"reviewIds": [
"r-91",
"r-88"
]
}
],
"disclaimer": "Drafts only — nothing has been posted, published or submitted anywhere. No guest's account of their own experience is disputed, and no refund, discount, upgrade or compensation is offered."
}Requirements
Integrations
No integrations required.
Before you start
- The reviews
- Optionally, past replies so drafts match your voice
- Requires the Starter plan or higher
Use cases
A backlog of reviews
Answered in one pass, each one specifically.
A bad week
Replies that do not sound defensive, written when you are not.
Operational review
What guests keep raising, ranked.
Pricing
Starter
Billed monthly through your PROMIVO subscription. Cancel at any time.
Runs consume your plan allowance for agent executions and tokens. See plan limits
FAQ
Will it post the replies?
No. It drafts. A reply published under your name is permanent and public, and that decision stays with a person.
What about an unfair or false review?
It drafts a measured reply and flags it for you. It will not publicly dispute a guest's account — that is the reply future guests judge you on, and it rarely reads the way the author intended.
Can it offer a guest something to fix it?
Never. It can say you would like to hear more directly, which invites the conversation without committing your money.
Reviews
Reviews open once customers have run this agent.
Need this agent to work differently?
Tell us what to change and our team will scope a customised version for your business.
Customize this agent