The check identifies the exact problem and a safe correction
A general score is not enough. The module shows the passage, the rule breached, the source and a possible direction for improvement.
Response to a guest complaint
The prepared response apologises and promises a refund, even though the hotel has not approved this decision.
Error, rule and next step
The system flags the unauthorised promise, suggests a neutral acknowledgement of the case and requires escalation.
- specific passage requiring change
- rule breached and knowledge source
- corrected version or mandatory escalation
One standard for AI and staff responses
The check can run before publication or audit a response that has already been published. Correction history helps monitor quality over time.
Checks facts
Compares claims and promises with approved knowledge and available sources.
Assesses communication
Checks the tone, length, specificity and required elements of the response.
Manages escalation
Blocks standard publication when the risk filter or a rule requires a human decision.
How does response control work?
Response read
The system receives a draft or published response together with the original comment.
Rules checked
AI compares the content with facts, brand policy, audit results and risk classification; it can also use Response Calibration.
Problem identified
The module highlights the passage, explains the breach and suggests a correction.
Record the outcome
Approval, correction or escalation is added to the response quality history.
Review requires statements, responses and rules
The assessment remains verifiable by linking the problematic passage to a specific rule or factual source.
What does AI use?
- source review or comment
- draft or published response
- approved knowledge and brand policy
- risk classification and optional Response Calibration
What does the team receive?
- review result against shared criteria
- identification of the error and its source
- corrected version or mandatory escalation
Review recommends but does not publish
The result helps the user improve the content or escalate the matter. The publication decision remains with a human until it can be safely automated in the Reputation Bot.
The user reviews the comments, edits or rejects the suggestion and approves publication themselves; high-risk matters are always escalated to a human.
The module does not publish responses, resolve disputes, approve compensation or bypass a block imposed by the risk filter.
Fewer errors and more consistent public communication
Employee and AI responses are checked against the same rules.
The user sees not only the assessment, but also the passage and suggested correction.
Further reviews reveal recurring problems and the effect of changes to the rules.
Review combines auditing, brand knowledge and risk filtering
Results can feed response suggestions, calibration and the escalation hub.
Related process elements
Frequently asked questions
Does the review work before publication?
Yes. It can check a draft before the user decides whether to publish it.
Can it audit responses that have already been published?
Yes. The result is added to the quality history and can help improve the rules.
Does it show the source of the error?
Yes. It links the problematic passage to a specific rule or a lack of supporting evidence.
Does it approve the response itself?
No. The user approves publication, while high-risk matters are always escalated to a human.