How to Configure and Use the Review Response Agent

The Review Response Agent helps businesses analyze customer reviews and generate context-aware responses using configurable AI workflows. It can analyze review details such as sentiment and other relevant attributes, use business and brand context to tailor the response, and route the generated response according to the workflow you configure.

Review response agents support different response approaches based on how much automation and human involvement you want:

Uses predefined templates to automatically respond to reviews.

Uses AI to analyze reviews and automatically generate and post unique, context-aware responses.

Uses AI to analyze reviews and generate context-aware reply suggestions that you can review and post manually.

Uses AI to analyze reviews, generate unique, context-aware responses, and send them through a human approval workflow before posting.

In this article, we'll focus on the Review response agent replying after human approval and walk through how to configure its workflow, review the available settings, test the agent, and publish it.

Click Review response agent replying after human approval to open the Agent configuration panel.

From the right panel, you can:

After making your changes, click Save.

The New review is received or updated trigger starts the workflow when a new review is received or an existing review is updated across your review sources and locations.

You can also add Trigger conditions to control when the agent should run. Click Add condition to define additional conditions for the trigger.

Click Save after configuring the trigger.

The Triage review action evaluates each incoming review to determine whether it requires a response and whether it should be treated as spam or irrelevant content.

In the Prompts tab, you can review the instructions used to evaluate the review and identify spam signals. The pre-configured prompt checks the Review Text against the Review Source and considers factors such as business-unrelated self-promotion, irrelevant content, unrelated contact information, scams or cryptocurrency promotions, lack of a genuine customer experience, and repetitive or automated content.

In the Fields tab, the action uses Review comment and Review rating as inputs and produces isSpam and spamReason as outputs.

The Context tab includes business information such as Category, Name, Sub Categories, Services, Products, and Description. The LLM model is set to Balanced.

The Based on condition branch checks the result of the Triage review action and routes the review to the appropriate workflow path.

In this agent, the branch separates reviews into two paths:

This branch determines which subsequent actions the agent executes based on the review's classification.

The Send an email alert action notifies the specified users when a review is identified as spam, so they can take the required action on the review site.

In the right panel, click the pencil icon next to the Send internal alert tool to configure the email.

Under Recipients, add the users who should receive the alert.

The Subject and Body fields are pre-populated with the review and spam details. You can edit them to customize the email content as needed.

Click Send to save the email configuration.

The Review analysis action analyzes non-spam reviews to identify the information and context needed to generate an appropriate response.

The action uses the review information provided by the workflow to analyze details such as language, topics, sentiment, staff or employee mentions, competitor mentions, and review severity.

The action is pre-configured with the required prompts, input fields, output fields, and business context. You can review these settings by switching between the available configuration tabs.

The Response generation action creates the final response to the review using the review details and the analysis generated in the previous step.

The pre-configured prompt guides the agent to tailor the response based on factors such as the review text, review rating, review severity, staff mentions, SEO, and other details identified during Review analysis. It also includes rules for handling different review scenarios, such as missing review comments, specific topics, staff mentions, and competitor references.

The Show response under approval action sends the generated review response through a human approval workflow before it is posted.

Click the pencil icon next to Handle response to configure how the response is handled.

Under Type of response, select how you want to generate the response:

Under Response handling, select how the response should be posted. For this agent, Post after approval is selected, so the response requires human approval before posting.

Under Select approval workflow, choose the approval workflow you want to use for reviewing the generated response.

Click Save to save the configuration.

After configuring the agent, run a test to verify that the workflow executes as expected.

Click Run test.

In Preview settings, select a review to use for testing. You can use the available filters to narrow the list of reviews.

Click Run test. The agent processes the selected review through the configured workflow and displays the results in Test details.

After the test completes, Test details displays the result of each step in the workflow.

A successful step is marked with a green check. You can expand and collapse the available sections under each step to review their outputs and details.

You can also expand View inputs to review the information provided to that step.

For example:

Review the test results to verify that the workflow produces the expected analysis and response. Click Exit test when you are done reviewing the results.

After configuring and testing the agent, you can choose how to save it.