# How To

# How to Access Governance

# Create Workspace

Only Tenant admins can create new Workspaces

1. Go to Admin Module > Workspaces
2. Click Create New and provide a name, description, and Image (optional)
3. You can add Members to the workspace
   

   

4. After you have added the users to the workspace, you can define their user policies
   

   

5. After you have added all the members, you can click on submit, and the workspace will be created

Create Users
Only Tenant Admins can create new users

1. Navigate to the Admin Console > User Management
2. Click on Create User and fill in the details
3. Assign the user to one or more workspaces from the dropdown. All the workspaces created by you will appear in this dropdown
4. Assign a User Policy (e.g., GenAI User, Workspace Admin)
5. Click Submit to onboard the user to the platform

# Manage Workspaces

1. From the Admin Console, select Workspaces
2. View existing workspaces along with their icon, their description, their members, owners, and modification date
   

   

3. You can search for any specific workspace by using the search bar
4. Click on the workspace that you want to manage
5. You can view the current members added to the workspace and their respective user policies
6. If you want to make changes to the workspace, click on the edit button
7. You can now manage the different user policies assigned to the members and click on save, post making the changes

# Manage Users

1. You can manage members and the policies assigned to them by clicking on the User Management
2. A complete list of all the members onboarded on the platform appears
3. Click on any member to view the workspaces they are a part of, and the respective user policies are assigned within that workspace
4. To make any edits for the member, click on Edit Account
5. The tenant admin can now make changes to the workspaces and user policies assigned to the member
6. The tenant admin can also reset the password for the user by clicking on ‘Reset Password’

Manage Asset Visibility

1. Inside the Expert Agent Studio, locate the agent for which you want to change the visibility
2. Click on the three-dot icon inside the agent card, and click on the Agent Profile to view the profile of the selected agent. You can now select the visibility of the agent from -
   • Private – Visible only to the creator
   • Public – Visible to all users in the workspace
     

     

Manage Toxicity Guardrails
Crosslinked with the Add-ons section within the Expert Agent Studio

# Monitor Agents

# Overview

1. Monitoring of agents can be done via the governance module
2. Inside the Governance module, a centralized view of all the agents within that workspace is visible
3. View a grid showing all agents with details like:
   • Agent Name, Type, Status, Transactions, Owner, Version
4. You can use filters to search by agent type or agent name
5. Customize visible columns to prioritize relevant data

# Key Fields in the Grid:

• Agent Name: The unique, user-assigned identifier of the agent
• Agent Type: Classification, such as Conversational or Automation
• Transactions: Total number of queries asked.
• Status: Indicates whether the agent is currently Active or Inactive
• Version: Current deployment version of the agent
• Owner: The user who created the agent

Search, Filter, and View Controls - To manage large-scale agent deployments efficiently:

• You can use the search bar, which will help quickly locate agents by name or type
• You can use filters to narrow down by status, agent type, or owner for targeted monitoring
• You can select the visibility of data columns in the grid according to your requirement
• You can also rearrange the column layout for personalized views

Agents can be activated or inactivated directly from this Module

Active agents are operational and available for use. To activate an agent -

• Click on the three-dot icon beside the agent name
• Click on settings
• Click on the Activate Agent button to activate the agent
  

  

  
  Deactivating an agent pauses it without deletion—useful for debugging, freezing usage, or managing cost. To deactivate the agent, you can simply unselect the activate button in the settings.

Clicking on any agent entry opens the Agent Details Page, providing deeper insights into its operation.

Overview Tab Includes:

• Total Transactions: Count of all prompt-response exchanges
• Model Cost: The Cumulative cost incurred based on model usage
• Input Tokens / Completion Tokens: Tokenized breakdown of input prompts and LLM-generated outputs
• Cognitive Processing Time: Total time taken by LLM to process the request
• Total Processing Time
• Transactions by Status

The Transactions Tab includes

• Lists every individual transaction for a transaction-level view
• Status of the transactions
• The tokens used, the model cost incurred, the processing time taken by the agent, and the user who initiated the transaction
  

  

# Monitor – Automation Agent Transactions

1. Click on any Automation Agent to monitor the performance
2. The overview of their performance and the transaction logs can be viewed
   

   

3. By clicking on the transactions tab, you can view all the transactions created
   

   

4. You can create transactions for automation agents from this page to test the behavior of the agent post publishing by clicking on ‘Create Transaction’. The create transaction button will only be enabled when the Agent has been activated in the Governance Module
5. Click on Create Transaction to test the published and activated agent
   

   

6. You can browse and upload files here for ‘File as an Input’ agents, or Type in your query for ‘Text as an Input’ agents
   

   

7. Post you have uploaded your files, you can click on Create
8. Post the transaction has been completed you can view the details
   

   

Located at the top of the view, this provides a quick summary of the task execution -

Input Panel has the following elements -

This makes it easy for users to visually confirm that the extraction was correct, without reading raw JSON.

Output Panel - This is a technical JSON representation of the result and supporting metadata
Traces Button - You can click on this to view the steps the agent performed to generate the output. The traces can be viewed at a timeline level, detailing the step-by-step process, or at a complete level, which just has the input and the agent output

You can download the output in a JSON format for further usage

Monitor – Conversation Messages

1. From the centralized view, Filter agents by type = Conversational Agent.

2. Click on any agent to view:

   • Number of messages
   • Input tokens, completion tokens, Messages vs time, and Users vs time
   • User satisfaction (thumbs up/down if enabled)
   • Cost metrics (token usage, model cost)
     

     

3. Click on the messages tab to view all the transactions and get details on a message level
   

   

4. At the top, you’ll see a compact summary of the session:
   

   

5. Timeline / Complete
These tabs let you switch between execution trace and full response view.

Timeline Tab (Default) - This is a visual representation of the interaction’s structure and duration.

# Complete Tab

This shows the final output or answer returned by the agent. Useful if you want to see only the result without tracing internal execution

You can also click on the chat icon beside the number of messages in the chat to view all the messages that were a part of the same chat session

# Monitor – Automated Workflow Transactions

1. Inside the Governance module, apply a filter for Automated Workflow
2. Click the agent to view the overall performance and a transaction-level performance view
   

   

3. Analyze the metrics such as Total transactions, Model Cost, Input tokens, Completion tokens, Cognitive processing time, and Total processing time
4. Click on the transactions tab to get a view of all the transactions made
5. Further, you can click on any particular transaction to view the details at a granular level