Shared Agent vs. Personal Copy: Making the Right Choice
When you discover an agent created by someone else, you have two options: use the shared agent as-is, or duplicate it to create your own personal version. Understanding when to choose each approach is crucial.
The Power of Shared Agents
One of the most powerful aspects of elvex is having a single agent that everyone in your organization uses. Shared agents provide:
Consistent experience across your team - Everyone gets the same responses, follows the same processes, and uses the same capabilities
Centralized improvements - When the agent creator updates the rules or adds new datasources, everyone benefits immediately
Organizational knowledge - Shared datasources mean the agent has access to company-wide information that helps everyone
Reduced maintenance - One person manages the agent configuration instead of everyone maintaining their own version
Use a shared agent when:
You want the standard, organization-approved version
The agent's purpose and rules align with your needs
You're comfortable with the AI model and datasources the creator selected
You want to benefit from updates the creator makes over time
When to Duplicate and Customize
Sometimes you need an agent tailored to your specific workflow or preferences. Duplicating creates your own independent copy that you can modify without affecting others.
Duplicate the agent when:
You want to experiment with different rules or instructions
You need to add personal datasources that aren't relevant to other users
You prefer a different AI model or provider
You want to customize the agent's behavior for your specific use case
You need capabilities beyond what the shared version offers
Important: Adding personal datasources to a shared agent changes the experience for everyone using that agent. If you want to add your own datasources, duplicate the agent first to make it your own.
To duplicate an agent, click the three-dot menu on the agent card and select "Duplicate." This creates a personal copy you can modify freely.
The Shared Agent Model
Think of a shared agent like inheriting a well-designed recipe. The creator has already:
Defined the agent's purpose and capabilities
Written the instructions that guide its behavior
Connected datasources that will help the agent perform its tasks
Chosen an AI provider
Decided what Slack workspaces the agent can be accessed from
However, just like you need your own ingredients and kitchen tools to make that recipe, you need to connect your own Actions to make the agent work with your personal accounts and data.
Datasources vs Actions: What's the Difference?
Understanding the distinction between datasources and Actions is crucial for properly using any shared agent:
Datasources (Often Shared):
Information repositories like company knowledge bases, documentation, or public data
Usually connected by the agent creator for the agent to use no matter who is using the agent
You typically don't need to reconnect these unless you want to add your own additional data sources
Actions (Always Personal):
Connections to your personal accounts and services (Gmail, Slack, Jira, Outlook, etc.)
Allow the agent to perform tasks on your behalf
Must be connected by each individual user
Cannot be shared for security and privacy reasons
Your First Step: Identify What Needs Personal Connection
Before using any shared agent, start by having a conversation with it to understand what it's designed to do. Ask questions like:
"What can you help me with?"
"What integrations do you need me to connect?"
"What Actions are required for you to work properly?"
The agent will typically tell you exactly what connections it needs and guide you through the setup process.
Reviewing Agent Configuration
You can always review how an agent is configured to better understand its capabilities and settings. To view the configuration:
Open the agent
Click the settings or configuration icon
Review the key settings, including:
Rules and instructions - See what guidelines the agent follows
Connected datasources - Understand what information the agent can access
AI model - Know which AI provider and model powers the agent
Enable personal context - Check whether this setting is enabled
Understanding "Enable Personal Context"
One important setting to check is Enable Personal Context. When this is enabled, the agent can access your personal information and conversation history to provide more personalized responses. When disabled, the agent treats each conversation independently without accessing your personal context.
If the agent's responses seem too generic or it's not remembering information about you, check whether "Enable Personal Context" is turned on. This setting affects your experience with the agent but doesn't change how others experience it.
When You Need to Connect Your Own Actions
You'll need to connect your personal Actions when the agent is designed to:
Send emails on your behalf
Create or update items in your project management tools (Jira, Asana, etc.)
Access your personal files or documents
Post messages to Slack channels (this is different than accessing an agent from Slack)
Interact with any service that requires your personal authentication
Key principle: If the agent needs to act as you or access your personal accounts, you must connect those Actions yourself.
Connect Your Actions
Open the Actions Directory by clicking on the > symbol in the top-right corner
Select the appropriate Actions
Try asking the agent to perform a task - it will tell you if it can't access required services
Remember that when you connect your personal Actions to a shared agent:
You maintain full control over your connected accounts
The agent creator cannot access your personal data or accounts
You can disconnect Actions at any time through your integration settings
Your personal information remains private and secure
Understanding these concepts will help you effectively use any shared agent while maintaining security and getting the results you expect.
