Every O-mega agent is highly configurable, with settings that control everything from their basic identity to advanced behavioral constraints. Understanding these configuration options helps you create agents that work exactly the way you need them to.
All configuration happens through your agent's Info section—a single location where every aspect of your agent's setup is accessible. Whether you're making a quick name change or overhauling your agent's entire approach to work, this is where you'll do it.
Identity
The identity settings establish who your agent is. These are the foundational elements that appear throughout the interface and influence how your agent represents itself.
Identity settings include:
- Name - Your agent's display name, used in conversations and across your workspace
- Title - A professional job title or role descriptor (e.g., "Marketing Assistant", "Research Analyst")
- Profile Image - An AI-generated or custom avatar that visually represents your agent
These elements are primarily about recognition and branding. They help you identify which agent you're working with and give the agent a consistent identity across all interactions.
Behavior
Behavioral settings control how your agent approaches work. These go deeper than identity, affecting the actual decisions and outputs your agent produces.
Behavioral configuration includes:
- Role Description - Core responsibilities that guide what your agent does and how it prioritizes
- Personality - Communication style and traits that influence how your agent interacts with you
- Rules - Hard constraints your agent must always follow, regardless of other instructions
- Learnings - Knowledge and preferences accumulated over time through your interactions
These settings work together to shape a coherent behavior profile. The role description defines the job; personality shapes the approach; rules set boundaries; learnings add personalization.
AI Model
Model settings determine which AI powers your agent's thinking. Different models have different strengths, and you can optimize for your specific use case.
Model configuration includes:
- Model - The primary AI model powering your agent's responses and reasoning
- Browser Automation Model - Optionally, a different model specifically for web navigation tasks
Default models are selected based on your subscription plan and work well for most use cases. You only need to change these if you have specific requirements around speed, capability, or particular task types.
Accounts
Account settings control your agent's access to external platforms. These are the credentials your agent uses when it needs to log into websites or access services.
Account configuration includes:
- Connected Accounts - Platform credentials like LinkedIn, Gmail, or other services
- Website Usage - Rules about which sites each account can be used on
- Primary Email - The email used when your agent creates new accounts on other platforms
Accounts enable your agent to work on your behalf across the web. Without connected accounts, your agent can still browse public content but can't access anything that requires authentication.
Tools
Beyond browser access, your agents can connect to external tools through integrations. These expand what your agent can do beyond web browsing and computer sessions.
Tool integration features:
- Connected via OAuth when you authorize access to external services
- Managed in the Tools section of your agent's configuration
- Each agent can have different tools connected based on their responsibilities
Tools include things like email services, CRM platforms, productivity apps, and other services that offer API integrations.
Best Practices
Effective configuration comes from understanding what matters for your specific use case and focusing there.
Start simple. Begin with name, title, and a basic role description. Add personality traits, rules, and other refinements as you learn what your agent needs. Over-configuration at the start often creates friction without benefit.
Be specific in rules. Vague rules are hard for agents to follow consistently. "Be careful with customer data" is less useful than "Never share customer email addresses outside of authorized communications." Write rules as clear, actionable constraints.
Review learnings periodically. Your agent accumulates understanding over time. Check what it has learned occasionally to ensure the accumulated context is accurate and still relevant.
Duplicating Agents
When creating new agents, you can use an existing agent as a starting point. Duplication copies configuration to give you a head start.
Duplicated elements include:
- Identity information (name and title, which you'll want to change)
- Role description and personality
- Rules and learnings
- Connected accounts (optionally, if you want the new agent to have the same access)
This is particularly useful when creating variations of an existing agent or when establishing a team of agents with similar base configurations.
Related: Choosing AI Models | Agent Personality and Rules