The Review Response Agent analyzes incoming reviews for sentiment, mentions, intent, and business context to generate timely, empathetic, and brand-aligned replies. It automatically identifies relevant mentions, extracts custom tokens, such as employee or location names, and drafts tailored responses that closely resonate with the original review. With optional tone and compliance controls, as well as human-in-the-loop approvals, the agent ensures consistency, speed, and quality across every reply while reducing manual effort and operational overhead.

Navigation

Go to ‘Settings,’ click ‘AI agents,’ and select ‘Review response agent.’

Goals

The Review Response Agent is designed to automatically generate timely, empathetic, and brand-aligned replies to customer reviews across all platforms. This helps improve customer satisfaction, strengthen brand reputation, and increase operational efficiency.

The predefined goals of this agent are:

1. Increase review coverage – respond to more reviews effortlessly across all platforms.
2. Save time and reduce costs – automate consistent responses to reduce manual effort and operational overhead.

Trigger

In Trigger, select when the Review Response Agent should run. ‘Created, Updated, Created & Updated' refer to the customer review event:

• Created – agent runs when a new review is posted.
• Updated – agent runs when an existing review is edited/updated.
• Created/Updated – agent runs for both events.

You can further refine triggers by clicking ‘Add conditions.’

For example:

• Rating – run the agent only for reviews with specific star ratings (e.g., 3, 4, 5).
• Source – run the agent only for reviews coming from specific platforms.

You can apply both conditions together (Rating + Source) to narrow down when the agent should respond.

Hover over the condition and click the bin icon to delete the condition.

Tasks

Switch to the ‘Tasks’ tab and configure the tasks the agent will perform.

Identify relevant mentions

The agent analyzes the review text to detect specific products or services mentioned.

• If multiple services or products are referenced, the agent prioritizes them based on their relevance to the overall review.
• These mentions are then used in responses, making them more personalized and context-aware while also helping improve SEO value.

You can click 'Add prompt' to provide specific instructions that guide how mentions should be prioritized or used in responses.

Enter the prompt and click ‘Add.’

Extract custom token references from reviews

The agent identifies custom tokens, such as employee names, location names, cities, or other custom location data in reviews. These tokens are used in the response to add personalization and make replies more authentic.

Add prompt:

You can click Add prompt to customize how tokens should be extracted and applied.

• Example: Always include employee names in the response if mentioned in the review.

Generate response based on analysis output and context

The agent analyzes review sentiment, tone, and context to draft tailored responses. It ensures empathy in negative reviews, appreciation in positive ones, and maintains a natural brand voice without repetition. Responses may also include relevant custom tokens for personalization.

Add prompt:

Click 'Add prompt' to guide how responses should be framed. You can also notice that prompts are added in the image below, which define how responses should be structured and refined.

Set response for approval or suggestion

You can configure the agent to send generated responses for approval before posting, or save them as suggestions.

• For reviews rated below 3 or with negative sentiment, responses are routed through the Response approval tool.
• For reviews outside the approval criteria, the generated response is added as a suggestion.

Click 'Response approval' to open the approval settings.

Select the desired approval workflow. Once selected, click ‘Save’ to finalize the workflow.

To modify the rule, click the Edit (pencil) icon and adjust the conditions as needed.

Run Test

Click the Run test option. It provides a real-time preview of how the Review Response Agent will work after you’ve configured tasks. It simulates the full process — from identifying mentions and extracting tokens to generating and routing responses — so you can validate the setup before publishing.

Identify relevant mentions – The agent highlights key mentions from the review (e.g., “jardín”) to guide a context-aware response.

Extract custom token references – The agent pulls custom tokens, such as business name and location, from the review.

Generate response based on analysis output and context – A tailored response is created, showing details, such as tone, length, content sources, response behavior, and a preview of the actual response generated.

Set response for approval or suggestion – The agent routes the generated response either for approval (based on your criteria, such as low ratings or negative sentiment) or adds it as a suggestion for positive reviews.

Approvals in waiting can be tracked under the Awaiting Approval tab within the Reviews AI module.

After reviewing the results, if you’re satisfied, pause the agent first.

After pausing the agent, the ‘Publish’ button becomes visible. Click it to publish the agent.

Outcomes

Outcomes measure how well the agent is performing against the goals.

Reviews responded: Total number of customer reviews the agent has responded to within the selected time period.

Review response rate: Includes responses published by the agent, not manual or suggested ones.

Time saved: Estimated time saved by automating responses compared to replying manually.

Configure Time Saved

The Configure time saved setting allows you to estimate how much time and cost your team saves by using the Review Response Agent to automatically reply to reviews.

Click the customized button right next to the ‘Time Saved’ section.

You can configure two fields here:

Time saved with agent responses for every message – Enter the average time (in minutes) it would normally take a team member to manually respond to a single review.

Hourly employee wage – Enter the average hourly rate (in your preferred currency) that your employees are paid.

Once these values are entered, Birdeye automatically calculates the total time and cost savings based on the number of reviews responded to by the AI agent. This helps businesses understand the real value and efficiency gained from automating their review responses.