AI Analytics & Agents

AI Agents extend the capabilities of Revel Digital's AI suite by enabling automated, recurring tasks that monitor and optimize your digital signage network. By combining AI intelligence with Scheduled Tasks, agents can run at regular intervals — analyzing network health, evaluating media performance, generating reports, and taking corrective action — all without manual intervention.

Overview

An AI Agent is a scheduled task configured with the AI Prompt task type. When the task runs, the platform executes the provided prompt against your account data using the same AI capabilities available in the Account Assistant. The key difference is that agents operate autonomously on a schedule, making them ideal for ongoing monitoring and optimization workflows.

Common use cases include:

• Network Monitoring — Detect offline devices, connectivity issues, or hardware anomalies and receive alerts or automated remediation.
• Media Performance (DooH) — Analyze play logs, audience impressions, and engagement metrics to evaluate content effectiveness across your network.
• Content Management — Automatically refresh playlists based on time of year, audience demographics, or performance data.
• Compliance Auditing — Verify that scheduled content is playing as expected and flag any discrepancies.
• Custom Workflows — Define virtually any recurring task using a custom prompt tailored to your specific operational needs.

Creating an AI Agent Task

AI Agent tasks are created through the Scheduled Tasks interface. For general information on creating and managing scheduled tasks, refer to the Scheduled Tasks article.

To create a new AI Agent task:

1. Navigate to Scheduled Tasks from the main configuration panel.
2. Click New Task.
3. Select AI Prompt as the task type.
4. Enter your prompt in the provided text field. This is the instruction the AI will execute each time the task runs.
5. Configure the schedule using a cron expression or the scheduling helper to define how frequently the task should run.
6. Click Save to create the task.

Writing Effective Prompts

The prompt is the most important part of an AI Agent task. It determines what the agent will analyze and what actions it will take. Here are some guidelines:

Be specific about the scope. Tell the agent exactly what data to examine and what criteria to evaluate.

"Review all devices that have been offline for more than 2 hours and send an alert summary via email."

Define the expected action. Specify whether the agent should report findings, make changes, or both.

"Analyze play logs for the past 24 hours across all retail location devices. Identify any playlists with less than 80% play completion and generate a summary report."

Include guardrails. Set boundaries on what the agent should and shouldn't do.

"Check for upcoming holidays in the next 7 days and suggest updated lobby content. Do not make any schedule changes — only provide recommendations."

Example Agent Prompts

Network Health Monitor

"Identify any devices that have gone offline in the past hour or are reporting CPU usage above 90%. Summarize the findings grouped by device group."

DooH Performance Analyst

"Analyze audience impression data for the past week across all advertising displays. Compare actual impressions against scheduled targets and flag any displays performing below 70% of their target."

Content Freshness Check

"Review all active playlists and identify any media items that have been in rotation for more than 30 days without being updated. Provide a list organized by playlist name."

Seasonal Content Agent

"Check for upcoming holidays and seasonal events in the next 14 days. Update the lobby playlist with appropriate seasonal imagery and messaging."

Task Notifications

When an AI Agent task completes, the account owner and any additionally configured recipients will receive an email summarizing the results. This email includes a detailed breakdown of the analysis performed and any actions taken by the agent during execution. This ensures full visibility into agent activity without needing to log into the platform after every task run.

Email notifications are especially useful for monitoring-focused agents where timely awareness of issues — such as offline devices or underperforming content — is important. They also serve as an audit trail for agents that make changes to your account, providing a clear record of what was modified and why.

Security & Permissions

AI Agents execute actions within your account, so controlling what an agent can access and modify is critical. Revel Digital provides fine-grained security controls through a dedicated AgenticAI role.

The AgenticAI Role

The AgenticAI role is a custom role specifically designed for controlling AI agent access to your account data. This role is assigned to the AI system and determines the boundaries of what any AI agent task can do.

To configure the AgenticAI role:

1. Navigate to Account Settings > Users & Permissions.
2. Locate the AgenticAI role.
3. Customize the permissions to match your desired level of access.

Permission Options

The AgenticAI role supports three permission levels for each resource type:

These permissions apply across all resource types in the platform, including devices, media, playlists, schedules, and templates. This allows you to create a precise security profile — for example, granting an agent view access to devices and play logs for monitoring purposes while restricting any ability to edit or delete content.

Media Approval

For organizations that require oversight of AI-generated content, the Media Approval feature provides an additional layer of control. When media approval is enabled, any media created or modified by an AI agent will not be scheduled for playback until it has been reviewed and approved by an authorized user.

This is particularly valuable for:

• Brand Compliance — Ensure all AI-generated imagery and content meets brand guidelines before going live.
• Regulatory Environments — Maintain human oversight of content in industries with strict advertising or content regulations.
• Staged Rollouts — Review agent-recommended changes before they reach your displays.

To enable media approval, navigate to Account Settings and enable the Media Approval option. Users with the Approver role will then be able to review and approve or reject any pending media before it enters rotation.

Security Best Practices

• Start with minimal permissions. Configure the AgenticAI role with view-only access initially, then expand permissions as you build confidence in your agent prompts.
• Use media approval for content-modifying agents. Any agent that creates or updates media content should operate under media approval to maintain human oversight.
• Test prompts manually first. Before scheduling an agent task, run the prompt through the Account Assistant to verify it produces the expected results.
• Monitor task history. Review the task run history regularly to ensure agents are operating as intended. Each task run logs its execution time and status.
• Separate monitoring from action. Consider using separate agent tasks for analysis/reporting and for taking action. This makes it easier to audit what changes are being made and why.