How to Configure and Use the Brand Compliance Agent

Summary 

We provide a comprehensive guide to building, testing, and running your Brand Compliance Agent. This covers initial setup, version management, review queues, and upload automation.

Who?

  • DAM administrators and brand managers configuring the system.

Why? 

Setting up the agent correctly ensures accurate compliance checks, automates the review queue, and integrates directly into your taxonomy and approval workflows.

How?

Complete the Prerequisites

  • Set up a compliance metaproperty. In your taxonomy, create a new metaproperty as a single-select field with four options: Compliant, Non Compliant, Partially Compliant, and Not Applicable.

  • Have your brand guidelines ready to convert into objective rules.

  • Check your permissions. You need Manage Agents to build/edit rulesets and approve status proposals, and Run Agents to execute checks on assets.

Create your Agent

  1. Navigate to the Agent Library via Settings > AI Agents > Manage AI Agents.

  2. Click Create agent and select Brand Compliance.

  3. Add your categories and rules, derived from your brand guidelines and regulations. Keep each rule precise and objective.

  4. Select the metaproperty you set up in Prerequisites to map the agent's output.

  5. Add a Name and Description, then click Create Agent.

Keep in mind:

  • You can reference one rule from another within the same category, but not across categories. A color definition set up in one category can't be referenced by a rule in a different category, repeat the definition where you need it.
  • The agent respects your Global AI Instructions, so compliance checks stay aligned with your brand tone and terminology.
  • Group related rules into one category rather than spreading them thin, it keeps cross-referencing simple and your ruleset easy to maintain.

Add a Reference Asset to a Rule

If a rule needs a visual reference, such as your logo file, attach it:

  1. Select an asset under Attachment in the right-hand menu.
  2. Edit the rule and type / to invoke the reference file.

Each rule can reference one asset. A reference file sharpens the accuracy of any rule that depends on it.

Test and Preview Rules 

Preview images and documents from your DAM while you build your rule set so you know exactly what your agent checks before deployment. Setup and testing do not consume credits, and no data is stored or saved during the process.

  1. Select Preview asset in the bottom right menu.

  2. Select Test category.

  3. Choose an asset to view a full-size preview.

  4. Run checks category by category to isolate specific rules. The interface displays feedback underneath the rule and its output.

  5. Review technical data points if your rules evaluate technical metadata.

  6. Edit the rule inline based on the feedback.

  7. Update and run the category with the same asset again to retest instantly without navigating away.

  8. Click the three-dot menu on the preview to replace or delete the test asset.

  9. Click Save when the rule evaluates correctly.

Manage Agent Versions and History 

We provide a full audit trail so your team can coordinate changes, trace rule updates, and troubleshoot effectively. The version history allows you to view past edits, compare changes, and restore known-good rulesets without destructive overwrites.

  1. Navigate to the Edit Agent page via Settings > AI Agents > Manage AI Agents.

  2. Open the version history panel to view all saved versions.

    • View the timestamp, the user who made the edit, and an indicator of what changed.

  3. Select any version to compare it against your current setup.

    • The version comparison highlights exactly what was added and removed at the line and character level.

    • Note: The comparison isolates the rules text only. Agent titles, descriptions, configurations, and auto-apply settings are ignored.

  4. Restore a historical version to use as your working baseline.

    • Click to load the selected version into the prompt editor.

    • Save the changes to commit this as the newest active version. This preserves a clean, chronological timeline without overwriting previous work.

Set your Agent's Status

Status controls who can see and run your agent, this is the same three-state model used across every agent type in the Agent Library:

  • Draft: visible only to users with the Manage Agents permission. Test your rules against a sample of assets before anyone else sees them.
  • Available: visible to users with the Run Agents permission, so your wider team can run it.
  • Disabled: not available for anyone to run. 

Run the Agent

  1. Go to the Asset Bank or a Collection. Filter down to the assets you want checked.
  2. In the bottom action menu, select Agents, then choose your agent.
  3. Maximum 1,000 assets per run.

Review your Results

A run doesn't change your taxonomy on its own, unless Auto-apply is enabled (see Automate the Agent below). The agent only proposes a compliance status per asset; someone with Manage Agents permission reviews it before anything updates, and has the final say.

  1. Open a completed run from the Activity screen. Each asset shows its proposed compliance status.
  2. Open the report on any asset to see the rule-by-rule breakdown, which rules passed, which failed, and why.
  3. If you agree with the agent, approve it. If you don't, override it with your own status, your decision always determines the asset's final compliance status, not the agent's proposal.
  4. Approve assets one at a time, or select a group and approve in bulk, one action writes the compliance status across every asset you selected. 
  5. Only once approved does the compliance value actually land on the asset and become visible and filterable elsewhere in the DAM, search, smart collections, and so on.

Two things stay in place after your decision, so you're never trading away visibility for control:

  • The original AI report doesn't disappear. Even if you override the agent's conclusion, its original report stays available on the asset's detail page, so anyone can see what the agent found versus what a human ultimately decided.
  • Overrides are auditable. Any compliance decision you change gets logged in the agent's History tab, giving you a full record of proposed-versus-final status for every asset.

Put Compliance to Work in your DAM

Once the agent is writing to your taxonomy, standard Bynder mechanisms turn that value into a real gate and review queue:

  • Content restrictions: the compliance metaproperty is just another taxonomy field, so you can attach permission sets to its values the same way you would for any restricted classification. An asset's visibility changes automatically as its compliance status changes.
  • Smart filters: build a saved smart filter or smart collection scoped to the compliance metaproperty so reviewers can pull their queue in one click. For example, "all Partially compliant" as an in-review queue, or "all Non-compliant" as the backlog that still needs action.

Automate the Agent

  • Trigger at upload: instead of running the agent manually, configure it to run automatically when new assets are uploaded, optionally scoped by metaproperty, tag, or file type. Compliance gets checked before the asset is available for wider use, not after the fact.
  • Auto-apply: skip manual sign-off entirely. The agent writes its verdict straight to the mapped metaproperty the moment a run finishes. This uses the same Auto-apply Approval User you configure in the AI Control Center, so an automated run still respects your organization's permission and access rules. See AI Control Center: Global AI Instructions.

AI can make mistakes. We recommend that you only apply automated compliance evaluations after careful consideration and in consultation with the respected teams in your organization (for example, consulting with your legal counsel on topics such as regulatory compliance).

Related Articles

Updated