# AI Recommender ## Introduction AI Recommender helps you identify license optimization opportunities in Microsoft 365 based on data already available in your Bsure Insights environment. It analyzes license assignments, sign-in history, subscriptions, prices, usage signals, and selected policy requirements to suggest where your organization may be able to reduce license costs by generating AI-powered recommendations. > AI Recommender provides decision support only. Bsure Insights does not automatically remove, downgrade, or change licenses. An administrator should always review recommendations before making license changes. We recommend at least 90 days of sign-in and usage history before relying on the recommendations. This is especially important for recommendations that evaluate whether Microsoft 365 E license holders may be better suited for Microsoft 365 F licenses, as user activity patterns may vary over time and sufficient history is required to ensure the recommendations are based on established usage patterns, not incomplete data. {% hint style="info" %} You can enable AI Recommender before enough sign-in history has been collected, but recommendations may not be generated until sufficient data is available and the better and more complete the data foundation is, the more useful the recommendations will be. {% endhint %} {% hint style="info" %} If your tenant has more than 20,000 active users, please contact Bsure Support before enabling AI Recommender. Bsure will help evaluate the recommended setup for your environment, as processing time and recommendation availability may vary for larger tenants. {% endhint %} ## How Bsure Selects Users for Recommendations Bsure evaluates users for AI recommendations only when they meet the required activity and licensing criteria. This helps ensure that recommendations are based on relevant usage patterns and do not include users who are disabled, unlicensed, or inactive. ### Who is included? Bsure evaluates users who meet all of the following criteria: | | | | ----------------------- | ------------------------------------------------------------------------------------------------------------------------------------ | | Active account | The user account must be enabled in Microsoft Entra ID. Disabled accounts are not evaluated | | Assigned licenses | The user must have at least one Microsoft 365 license assigned. Users without assigned licenses are excluded | | Recent sign-in activity | The user must have signed in within the defined activity period. Users without recent sign-in activity may be classified as inactive | ### New accounts Users holding licenses covered under [Supported Licenses by Recommendation Type](#supported-licenses-by-recommendation-type) who were created less than 60 days ago are excluded from scoring entirely. They receive no recommendation during this period. This prevents false downgrade or removal recommendations for newly onboarded users who have not yet had time to establish a pattern for their assigned licenses. ## Available features AI Recommender includes separate recommendation features that can be enabled depending on what is available for your tenant. Each feature evaluates a specific license optimization scenario and may have its own data requirements, policy criteria, and output views. ### Recommend removing unused premium licenses This feature identifies premium licenses with little or no active usage and highlights licenses that may be safe to remove. * Identify premium licenses with little or no active usage * Highlight licenses that may be safe to remove * Support license cleanup and cost optimization * Show potential savings where license cost data is available Premium license recommendations are shown in license optimization views and assessments ### Recommend F licenses for E license holders This feature evaluates whether users with Microsoft 365 E licenses may be better suited for Microsoft 365 F licenses based on usage patterns and policy criteria. The recommendation considers whether an F license can support the user’s current usage pattern without removing capabilities the user appears to need. #### Mailbox and desktop usage data This recommendation analyzes mailbox and OneDrive usage statistics from Microsoft 365 reports, together with desktop usage signals from sign-in history. Users may be excluded from these recommendations when mailbox or OneDrive usage exceeds F-license limits, or when desktop app usage indicates that capabilities beyond an F license may still be required. > If this usage data is missing or incomplete, recommendations may be delayed, reduced, or not generated for some users. #### Prerequisites To generate these recommendations, your tenant must have: * Both E licenses, such as Microsoft 365 E3, and F licenses, such as Microsoft 365 F3, available * Eligible Microsoft 365 E licenses assigned to users * At least 90 days of sign-in and usage history for Bsure to evaluate usage patterns * Identifiable user information available in Microsoft 365 reports. Please follow [this guide](https://learn.microsoft.com/en-us/troubleshoot/microsoft-365/admin/miscellaneous/reports-show-anonymous-user-name) {% hint style="info" %} For tenants within the supported size range, recommendations are normally expected to be processed within 24 hours. For tenants with more than 20,000 active users, processing may take longer, and Bsure cannot guarantee that recommendations will be completed within this timeframe. {% endhint %} {% hint style="info" %} These recommendations reqiure identifiable user information to be available in Microsoft 365 reports.\ See [Microsoft documentation](https://learn.microsoft.com/en-us/troubleshoot/microsoft-365/admin/miscellaneous/reports-show-anonymous-user-name) for how to disable concealed user information in reports. {% endhint %} #### Security Level Policy When enabling E-to-F recommendations, you must choose a Security level policy. This policy tells AI Recommender which minimum Entra security capability should be preserved when evaluating possible F-license recommendations. This helps ensure that recommendations preserve advanced Microsoft Entra security capabilities where they are required. | | | | ---------- | ------------------------------------------------------------------------------------------------------------------------------ | | No minimum | Use when E-to-F recommendations should not require a specific Entra P1 or P2 security level | | P1 minimum | Use when recommendations must preserve capabilities such as Conditional Access and other P1-related requirements | | P2 minimum | *Recommended default*. Use when recommendations must preserve advanced identity protection or risk-based security requirements | We recommend using **P2 minimum** as the default policy, unless your organization has explicitly decided that a lower security requirement is appropriate for the users being evaluated. {% hint style="info" %} For more information about Microsoft Entra security capabilities, see Microsoft’s official documentation for [Microsoft Entra Conditional Access](https://learn.microsoft.com/en-us/entra/identity/conditional-access/) and [Microsoft Entra ID Protection](https://learn.microsoft.com/en-us/entra/id-protection/overview-identity-protection). {% endhint %} #### When do E-to-F recommendations appear? These recommendations are generated after AI Recommender has been enabled, initialization has completed, and Bsure has collected enough sign-in and usage history to evaluate established user activity patterns. If your tenant was recently onboarded to Bsure, recommendations may not appear immediately, but they will be generated automatically once sufficient data is available. No additional action is required. Once generated, they will appear under **AI Recommendations** in the **My Team** and **My Organization** views and drilldowns. ## Enable recommendations To enable AI Recommender features, go to Settings -> System and the AI Recommender tab. The AI Recommender settings page shows which recommendation types are available for your tenant and displays the current configuration.

The settings page for AI Recommender

Use the toggle switch to turn on the recommendation features you want to enable. Changes are not applied immediately when you use the toggle switches. After selecting the recommendation features you want to enable or disable, click **Save** to apply the configuration.

Use the toggle switch to turn ON or OFF the feature

When enabling E-to-F recommendations, you must also choose a [Security level policy](#user-content-fn-1)[^1].

Select the Security level policy for E-to-F recommendations

### Apply changes When you click **Save**, Bsure displays a confirmation dialog with the AI Recommender configuration that will be applied. To confirm the configuration, click **Save changes**. Bsure then starts a re-initialization process for all enabled AI Recommender features, including database migrations, rule generation, and complexity analysis.

This is the Save modal that is displayed

### Processing steps After you enable or update AI Recommender, Bsure starts a processing workflow to prepare the recommender, analyze license and usage data, and generate updated recommendations. The process runs automatically, but it may take several minutes depending on the number of users, subscriptions, and enabled recommendation types. AI Recommender works in several steps: 1. **Initialization** prepares the required data, rules, and internal tables. 2. **Training** builds the recommendation foundation from license and usage data. 3. **Scoring** evaluates users against the trained recommendation rules. 4. **Publishing** makes recommendations available in the relevant dashboards and views. After saving the configuration, System Health may show AI Initialization or AI Training as In progress while Bsure Insights prepares and updates the recommendations. Recommendations may not appear immediately or may not reflect the latest configuration until the processing is complete. ## Notes ### Privacy And Data Handling AI Recommender is designed to run inside the customer-side Bsure Insights environment. The analysis is performed against data in the customer installation and customer data stores. Recommendations are generated from customer-side data such as license assignments, sign-in signals, usage reports, and configuration. Bsure Insights does not automatically perform license changes based on these recommendations. ### Costs Enabling AI Recommender may slightly increase the running cost of Bsure Insights because additional background jobs, database operations, and analysis processes are executed. > Enabling AI will increase running cost of Bsure Insights slightly but should be less than 100 USD per month. Please monitor cost after configuration changes in AI Recommender. For most customers, the additional cost is expected to be limited, but we recommend monitoring Azure costs after enabling AI Recommender, especially in environments with many users or many license combinations. ## Appendix ### Supported Licenses by Recommendation Type Lists the license scope currently supported by each AI Recommender feature. It explains which licenses can be evaluated for premium license removal and which Microsoft 365 E and F licenses are included when evaluating whether an E license holder may be better suited for an F license. {% file src="/files/r5rrmsDHECmHcucoaagJ" %} ### Frontline License Eligibility Checks The following appendix explains the eligibility checks Bsure performs before recommending that a Microsoft 365 E license holder may be better suited for a Microsoft 365 F license. These checks help ensure that the user’s current mailbox, OneDrive, and desktop app usage fit within Frontline license capabilities. {% file src="/files/frBYus7oRDYM6kgzcWwH" %} [^1]: The Security level policy defines the minimum Microsoft Entra security level that should be considered when generating E-to-F recommendations. --- # Agent Instructions: Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter: ``` GET https://docs.bsure.io/user-guides/settings/system/ai-recommender.md?ask= ``` The question should be specific, self-contained, and written in natural language. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.