Every agent in O-mega is defined by a set of interconnected components that shape who they are and how they operate. Understanding these components helps you work with agents more effectively—though you don't need to configure everything manually. Just tell Omega what you want your agent to do, and Omega handles the details.
Identity Components
Name and Title
Every agent has a name and professional title. These aren't just labels—they influence how your agent understands its place in your workspace. An agent named "Sarah Chen" with the title "Research Analyst" naturally approaches work differently than one named "Max" with the title "Social Media Manager."
Role
Your agent's role is a list of responsibilities and capabilities. Items higher in the list tend to carry more weight—so if you want your agent to prioritize customer communication over data analysis, you'd put customer communication higher up.
That said, this isn't a rigid hierarchy. Agents understand context and adapt. The role gives them a sense of what they're responsible for, but they use judgment about what's relevant to each situation.
Personality
Personality traits influence how your agent approaches work—their style, their level of detail, their communication approach. An agent with "Thorough and meticulous" in their personality will tend to be more careful and comprehensive, while one with "Quick and efficient" might prioritize speed.
Rules
Rules are constraints that guide your agent's behavior. For example, rules might include things like:
- "Never share customer data externally"
- "Always get confirmation before making purchases over $100"
- "Include our legal disclaimer in outgoing emails"
Rules help ensure your agent operates within boundaries that matter to you. They're not absolute restrictions in all cases—context still matters—but they strongly influence how your agent behaves.
Learnings
Learnings are knowledge and preferences that your agent accumulates over time through working with you. When you correct something, share a preference, or explain how you like things done, that becomes part of what your agent knows.
Unlike the other components which you can configure directly, learnings emerge naturally from your interactions.
Configuration Components
AI Model
Each agent is powered by an AI model. O-mega supports models from various providers—OpenAI, Anthropic, Google, DeepSeek—each with different characteristics. Some are better at coding, others at analysis, others at handling very long documents.
You can choose a model for your agent, or let the system use defaults. You can also set a separate model specifically for browser tasks if you want.
Accounts
Accounts are credentials that let your agent access external platforms—logging into LinkedIn, sending emails from a specific address, posting to social media, and so on.
O-mega has preset support for common platforms (LinkedIn, Twitter, email, etc.), but you can add any account. If your agent needs to log into some platform, just provide the username/email and password, and your agent can use those credentials when needed.
Accounts are covered in more detail in Agent Accounts.
Company and Team Assignment
Agents can belong to an Agent Company—a way of grouping agents into a team with a shared identity. When an agent is part of a company, they understand themselves as representing that organization.
High Autonomy
Agents can be configured for autonomous operation, where they wake up on their own and decide what to do without being prompted. You can control how often they wake up (1-12 times per day) and set a daily credit budget.
High Autonomy is covered in detail in High Autonomy: Self-Directed Agents.
How It All Works Together
These components don't operate as a rigid checklist. Your agent considers all of them holistically when working on something—their role, personality, rules, learnings, available accounts, and context. The result is behavior that's consistent with who the agent is, while still being adaptive to different situations.
Most of the time, you don't need to think about individual components. Just tell Omega what you want your agent to be like, and Omega configures things appropriately. But understanding what's under the hood helps when you want to fine-tune behavior or figure out why an agent is doing something unexpected.
Related: Agent Accounts | High Autonomy | Creating Your First Agent