Agent Library: Creating, Running, and Updating Agents

Summary

The Agent Library is the central database for your agents. Here you create new agents and edit, manage, or delete existing ones, with full visibility and control. Integrated directly with the Asset Bank and powered by large language models, agents use capabilities like object recognition and natural language generation to automate metadata enrichment, image transformation, and brand compliance verification.

Who?

• Target Audience: Bynder Administrators, Brand and Content Managers, Designers, and Digital Asset Managers.

• Permissions Required: Users need the Manage Agents permission to access the Agent Library.

How?

Accessing the Agent Library

1. Log in to your Bynder portal.

2. Go to Settings > AI Agents > Manage AI Agents to open the Agent Library.

   

Creating an AI Agent

1. Open the Agent Library.

2. Click the Create agent in the top-right corner. 

   

3. Select the type of agent you want to build from the setup prompt: 

   • Brand Compliance: Ensure consistency inside your DAM.
   • Enrichment: Automate taxonomy management across your assets.
   • Transformation: Use AI-powered editing tools for creative tasks.
   • Governance: Find out where your assets are used on the web.
4. Build your agent according to your needs:
   • Brand Compliance: Add categories and rules. See the Brand Compliance Agent article.
   • Enrichment: Enter a natural language prompt describing the exact task the agent should fulfill. See the Enrichment Agent article.
   • Transformation: Configure the transformation settings. See the Transformation Agent article.
   • Governance: No further setup required; you can customize its name. See the Governance Agent article.

5. (Auto)generate a Name and Description for your agent, then click Create Agent to save it to your library.

   

Running an AI Agent on Assets

1. Navigate to your Asset Bank.

2. Select the target assets you want the agent to process.

   

   • Constraint: You can select a maximum of 10,000 assets per run for Enrichment Agents. For other agent types, the maximum is 1,000 assets per run.

3. Click Select agent from the action toolbar. 

   

4. Select the desired agent, confirm, and run. A notification confirms that background processing has started.

   

5. Once processing completes, review the output before applying changes to your assets. See Workspace: Reviewing and Approving Agent Suggestions.

Running AI Agents using Bynder’s AI Automation Workflows?

• How To Use Bynder's AI Automation Workflows
   

Managing and Updating AI Agents

1. Navigate to the Agent Library.

2. Locate the agent you wish to adjust.

3. Use the management tools to edit prompts or delete agents entirely.

Agent Version History (Enrichment Agents)

When updating Enrichment Agents, administrators have access to a full version history log. This ensures team coordination and safe experimentation by allowing you to track exactly who changed a prompt, what was altered, and restore previous versions.

Note: This feature is currently available exclusively for Enrichment Agents.

Using the Agent Version History

When you open an Enrichment Agent in the Edit Agent page, a dedicated version history panel is available on the side of your screen.

• View Version History: The panel displays a chronological, timestamped log of every saved version, including the name of the administrator who made the edits.

• Compare Versions: Select any historical version from the log to view a side-by-side comparison with your current active prompt. Added text is highlighted in blue, and removed text is highlighted in red and strikethrough. We highly recommend using the comparison view before saving major changes to ensure accuracy.

• Restore a Previous Version: You can load a historical version back into your prompt editor to use as your working baseline. When you click Save, this restored text is committed as the newest active version at the top of your log. This preserves a clean, linear timeline without permanently overwriting historical data.

Important Considerations

Advanced Features

Agent status

Use the agent status dropdown to manage the visibility of your agents. This is helpful when an agent isn't ready for all users yet.

• Draft: only visible to users with the Manage Agents permission.
• Available: shown to all users.
• Disabled: not shown to any users.

Auto-Applying the Agent suggestions

Agent outputs go to the Agents activity queue for human review by default, since AI can make mistakes. If you've carefully reviewed an agent's accuracy across a diverse set of assets for a specific workflow and are confident in its results, you can turn on Auto-apply for that agent to skip manual verification going forward.

How it works: When enabled, the agent's suggestions skip the review stage and are applied to the asset's metadata automatically, without a person checking each one first.

How to enable it: Open the desired agent in the Agent Library, go to its settings panel, and switch the Auto-apply toggle to ON.

The safety net: Auto-apply removes the review step, not accountability. Every automated change is logged, so you can check what was updated or reverse an entire run at any time from the History tab in the Workspace. This lets you catch and undo mistakes after the fact, it isn't a guarantee the agent got everything right.

AI can make mistakes. We recommend that you only leverage auto-apply after careful consideration and in consultation with the respected teams in your organization.

Related Articles

Workspace: Agent Activity - Reviewing, Editing, and Approving Agent Suggestions

How To Use Bynder's AI Automation Workflows

Updated September 07, 2026 07:00