The AI Search Optimization Agent automates the analysis of your visibility, citation, sentiment, and accuracy reports and generates contextually relevant, prioritized recommendations to improve each metric. You can configure the agent to generate recommendations for selected locations or all locations on a recurring schedule based on insights from your AI Search reports.
The agent can generate recommendations across areas such as Content, FAQs, Listings, Reputation, Outreach, Accuracy, and General, and you can customize the agent's configuration to align recommendations with your business requirements.
Access the AI Search Optimization Agent
From the left navigation menu, go to AI Search.
Expand Agents.
Select Optimization Agent.
Switch to the Library tab.
Hover over Optimization agent and click Use agent.
Configure the AI Search Optimization Agent
Click the Optimization agent to open the Agent configuration panel.
From the right panel, you can:
Modify the Name of the agent.
Review the pre-configured Goals that define what the agent is intended to achieve.
Click the pencil icon next to Locations to select the locations where you want the agent to generate recommendations.
After making any changes, save the configuration before proceeding.
Set the schedule for recommendations
The Set schedule for recommendations trigger determines when the agent automatically runs to generate recommendations that can help improve your AI Search score.
The pre-configured agent is set to run Monthly. You can configure the following:
Date — The day of the month when the agent should run.
Time — The time when the agent should run.
Click Save after configuring the schedule.
Analyze AI Search reports
The Analyze AI Search reports action analyzes the selected AI Search reports to identify trends, anomalies, and gaps and create an insight summary. This analysis provides the information the agent uses to generate relevant recommendations.
Click the pencil icon next to the Analyze reports tool to choose which reports the agent should analyze.
Under Which reports you want to analyze, you can select:
Visibility reports
Citations reports
Accuracy reports
Sentiment reports
You can select individual reports or click Select all to include all available reports.
Click Apply to save your selection, then click Save on the action configuration.
The pre-configured agent includes all four report types, so the agent can use insights across visibility, citations, accuracy, and sentiment when generating recommendations.
Configure recommendations
The Configure recommendations action lets you choose the types and number of recommendations the agent generates.
Click the pencil icon next to the Configure recommendations tool to open its configuration.
Under "What type of recommendations you want to generate," select the recommendation categories you want the agent to generate. The available options in the pre-configured agent are Content, FAQ, Listings, Outreach, Accuracy, and General.
Under No of recommendations to be generated, select the number of recommendations to generate. You can select up to 10 recommendations.
The Select agent to generate FAQ content and Select agent to generate blog content fields let you choose the agents used when the corresponding recommendation requires generated content. The pre-configured setup already has agents selected for these fields.
The Recommendations field is pre-selected and passes the generated recommendations to the next step in the workflow.
Click Save after configuring the settings.
To learn more about the FAQ page and Blog page agent, read:
Generate recommendations
The Generate recommendations action converts the insights from the Analyze AI Search reports step into structured recommendations designed to improve your AI Search score.
In the Basic tab, you can modify the Action name and review the pre-configured description.
The action is already configured with the required prompts, input and output fields, context, and LLM model. You can switch between the Basic, Prompts, Fields, and Context tabs to review the complete configuration.
The Prompts tab defines how the insights are converted into recommendations. The configured instructions ensure that permitted recommendation text is refined for clarity while preserving the original meaning.
The Fields tab shows Analyze recommendations as the input and Final recommendations as the output.
The Context tab includes Business Website as additional context, and the LLM model is set to Balanced.
Send to AI Search
The Send to AI Search action sends the generated recommendations to the AI Search Recommendations section so they are available for review.
Click the pencil icon next to the Save recommendations tool to review the configuration.
The Recommendations field is mapped to Final recommendations from the previous step. The configuration also includes the agents used to generate content for relevant recommendation types, including:
FAQ agent — Uses the selected FAQ agent when an FAQ recommendation requires generated content.
Blog agent — Uses the selected blog agent when a content recommendation requires generated blog content.
Click Save to save the configuration.
The generated recommendations are then sent to the AI Search Recommendations section, where they can be reviewed and acted on.
Run a test
After configuring the AI Search Optimization Agent, run a test to verify that the workflow generates recommendations as expected for a specific location.
Click Run test.
In the Preview settings window, select the location you want to use for the test. The list displays the available locations.
Click the radio button for the location you want to test, then click Run test.
The agent runs the configured workflow for the selected location, allowing you to review the generated outputs before activating the agent.
Review the test results
After the test completes, the Test details panel displays the outcome of each step in the workflow.
Each completed step is marked with a green check. You can expand the sections under each step to review the inputs and outputs generated during the test.
For example:
Analyze AI Search reports — Expand Recommendations to review the data generated from the selected AI Search reports.
Configure recommendations — Expand Recommendations to inspect the recommendation configuration used in the test.
Generate recommendations — Expand Final recommendations to review the agent-generated recommendations.
The Final recommendations output can include details such as the recommendation title, summary, priority, impact type, expected impact, and associated theme.
After reviewing the test results, click Exit test to return to the agent configuration.
Activate the Agent
After configuring and testing the agent, you can choose how to save it.
Click Activate to publish the agent and make it available for use.
Click the dropdown arrow next to Activate and select Save as draft to save the configuration without publishing the agent.
Once the agent runs on the scheduled date and time, it generates recommendations based on the configured settings. You can view the generated recommendations by navigating to AI Search, expanding Human actions, and selecting Recommendations.
To learn more about reviewing and managing AI Search recommendations, read: How to Use AI Search Recommendations to Improve Visibility, Engagement, and Search Performance.