Deploying pre-built agents

Pre-built agents are ready-to-use AI solutions designed to handle specific tasks within an enterprise’s various departments. These agents can be deployed to automate workflows, improve efficiency, and streamline operations. They come pre-configured with core functionalities tailored to different business needs, allowing for seamless integration and execution with minimal setup.

This guide provides step-by-step instructions on how to deploy these pre-built agents.

Step 1: Access the agent store

  1. Navigate to the agents page: Go to the Agents section within your platform.

  2. Open the agent store: Click on the ‘Agent Store’ button located in the top-right corner of the page to browse available pre-built agents.

  1. Search for agents: Use the search bar to find agents that fit your specific use case by entering relevant keywords.

Step 2: Select the agent

  1. Choose the desired agent: Once you find the agent you want to deploy, click on its name to view more details.

  2. Deploy the agent: On the agent page, click the ‘Deploy Agent’ button to initiate the deployment process.

Step 3: Configure the agent

  1. Agent overview: You will be directed to the Agent Overview page, where you will configure the agent.

    • Agent name: Enter a unique name for your agent.

    • Agent description: Provide a brief description of the agent’s purpose and functionality.

  2. Set agent access:

    Select the appropriate access level for your agent:

    • Public agent: The agent link will be publicly accessible, enabling anyone with the link to view the agent dashboard and operate the agent.

    • Private agent: The agent link will be restricted to invited operators only, ensuring that only authorized users can view the agent dashboard and interact with the agent.

Sharing a public agent link

To share a public agent link, follow these steps:

  1. Open the Agent Dashboard.

  2. Navigate to the Output Section.

  3. Click the three-dot menu (⋮) in the top-right corner.

  4. Select Share Agent.

  5. A form will appear with the shareable agent link.

  6. Copy the link and share it with users who need access outside ZBrain.

Collecting user information

Steps to enable user info collection:

  1. In the Share Agent form, enable the ‘Collect User Info’ option.

  2. Select the required details to collect:

    • Name

    • Email ID

    • Phone Number

  3. Click ‘Update’ to save the settings.

Once enabled, users must enter their details before accessing the agent, enhancing tracking and engagement management. Configuring these settings allows you to control access while collecting valuable user data for better operational insights.

Adding operators to a private agent

To add operators and grant access to a private agent, follow these steps:

  1. Open the Agent Dashboard.

  2. Navigate to the Output Section.

  3. Click the three-dot menu (⋮) in the top-right corner.

  4. Select Add Operators to access the User Management page.

  5. Choose an access method:

    • Custom: Enter the operator’s name in the search field and select them.

    • Everyone: Grant access to all eligible users.

Managing operator permissions:

  • Click the three-dot menu (⋮) next to the operator’s name.

  • Select Transfer Ownership to assign a new owner.

  • Select Remove User to revoke access.

By following these steps, you can efficiently manage operator access and permissions for a private agent.

  1. Proceed to the next step: After filling in the details, click ‘Next’ to continue to the next configuration step.

Step 4: Define input sources

  1. Create a queue: In the Create a Queue page, you will define the input sources from which the agent will receive data and execute tasks in sequence.

  2. Add input sources:

    • Click the '+' symbol to add the necessary input sources.

    • Use the search bar to find and select the input sources relevant to your agent's tasks.

  • Some sources may require creating a new connection. Once the source is added, it will automatically trigger the agent in response to specific events occurring within the connected source.

  1. Set up sample input for testing

Test flow execution with sample input data to enhance testing precision during agent flow creation. To enable sample input testing:

  • Switch on 'Use input sample for testing' when creating or editing any output step

  • Select your preferred sample data type from the available options

Sample data types:

  • Text: Manually paste or type raw text to simulate user input

  • File: Upload supported files (JPEG, PNG, PDF, TXT, DOCX, XLSX) for automatic text content extraction

  • URL: Enter a web link for the flow to fetch and parse the page content

Flow integration:

  • Sample inputs automatically flow into the Catch Webhook step for use during the design and testing phases

  • Modify or disable input sample settings at any time before completing the agent configuration

Testing vs production

  • Sample inputs are used only during agent setup and testing

  • At runtime, agents respond strictly to real-time user inputs, and no test data is injected into live operations

  • This feature allows you to validate your agent's behavior without requiring external data sources during development

  1. Complete input source setup: After adding all required input sources, click ‘Next’ to proceed.

Step 5: Define the agent’s Flow

  1. Set the agent’s Flow: On the Define Flow page, customize the steps that the agent will follow during execution. The platform provides predefined steps using core elements and tools, but these can be adjusted based on your needs.

  2. Save and proceed: Once you have made the necessary customizations, click ‘Save’ to save the flow and then click ‘Next.’

Step 6: Additional settings

  1. Add output sources: On the Additional Settings page, you can add output sources where the agent will append its results:

    • Click the '+' symbol to add output sources.

    • Enter relevant details such as document IDs, sheet IDs, or page IDs, depending on the selected output source.

  1. Add instructions or additional information: Provide clear and concise instructions or additional details for the agent’s operation:

    • Keep instructions brief and to the point.

    • Use this space to specify any essential guidelines the agent should follow.

  2. Transfer output to other agents: If required, pass the agent’s output to another agent. Click ‘Add’ to select the agent that will handle the output. Only one agent can be added at a time for output transfer.

  1. Configure manual and automatic agent triggers:

  • Allow manual trigger

    • Toggle this setting to enable or disable manual triggering of the agent.

    • When enabled, the agent operator can initiate the agent manually by pressing a button.

    • Useful for testing, on-demand executions, or scenarios where agent actions should only occur with operator approval.

  • Auto trigger interval

    • Select the time interval at which the agent should automatically trigger.

    • This ensures periodic execution without manual intervention.

    • Options may include fixed intervals such as every 5 minutes, 10 minutes, or 30 minutes.

    • Helps in automating repetitive tasks or scheduled updates.

  1. Proceed: After finalizing the settings, click ‘Deploy Agent.’

Step 7: Monitoring and managing your agent

Once the agent is deployed, ZBrain equips you with a robust set of features to monitor, manage, and optimize its performance seamlessly:

1. Agent dashboard

Get full visibility into the agent’s activity:

  • Upload new files for processing using the ‘Upload’ button.

  • Search and view queued or processed documents.

  • Access detailed processing reports directly from the main viewing area.

  • Track activity trends (recent actions, error rates, throughput) and monitor overall agent performance.

2. Queue management

Efficiently manage your document pipeline with intuitive filtering:

  • Filter queue by status:

    • Show All

    • Processing

    • Completed

    • Failed

    • Pending

  • Click ‘Apply’ to quickly locate and manage specific items.

  1. Monitor feature Access comprehensive oversight of your AI agents with automated evaluation and performance tracking:

  • Expand the agent activity section to access session logs and step details

  • Click the 'Monitor' button located in the top-right corner of the step overview panel

  • Click 'Configure now' when prompted with 'Added for monitoring'

  • Set up monitor parameters on the redirected Events > Monitor page for continuous assess and response quality evaluation

4. Performance analytics

Gain deep insights into agent performance:

  • Time metrics: Monitor total processing time and average session duration to evaluate agent efficiency.

  • Quality scores: Track user satisfaction ratings to gauge response quality and effectiveness.

  • Resource usage: Analyze token consumption and associated costs to manage resource utilization.

  • Session logs: Review timestamped logs of each session for detailed insight into individual interactions.

4. API Integration

Seamlessly integrate the agent into external applications:

  • Use the provided Agent URL endpoint and authorization token for secure access.

  • Jumpstart development with sample code available in:

    • cURL

    • JavaScript

    • Python

  • Enable consistent, real-time interactions between your product and ZBrain.

Last updated