Topics Covered:

• Universal Workers: General-purpose, reusable AI agents that can be customized and shared
• Specialized Workers: Deeply customized agents built for specific use cases using Canvas/Studio
• Assistant Workers: Context-specific agents that exist only to assist humans in specific workflows
• Worker lifecycle management (creation, deployment, versioning, sharing)

Worker Types

Universal Workers

Universal Workers are general-purpose AI agents designed to be flexible, modular, and reusable across various tasks. They are created using the Universal Worker Builder, where Builders define:

• Profile: Name, avatar, tags, visibility (public/private)
• Knowledge: Vector memory sources, context documents, semantic search
• Brain: LLM configuration, behavior prompts, role, tone, output format
• Skills: API integrations and other embedded capabilities (via Connectors and Specialized Workers)

Universal Workers support OAuth integrations, file uploads, and dynamic session contexts. They are ideal for users who need consistent, intelligent support across a range of workflows.

Specialized Workers

Specialized Workers are workflow-driven agents built for specific, structured tasks using the Canvas interface. They rely on node-based logic, supporting complex control flows, conditions, and data operations. Key characteristics:

• Built visually: Drag-and-drop Canvas with node types (API calls, logic blocks, vector search, etc.)
• Composable: Can be used as "Skills" within Universal Workers
• Modular: Each node can use its own connector and execute independently
• Flexible: Suitable for automation, backend processes, and domain-specific logic

Assistant Workers

Assistant Workers are lightweight, contextual agents tailored to assist humans in narrow scenarios, like connecting to a user’s Gmail or calendar.

• Technically implemented as Universal Workers with preconfigured OAuth (personal) connectors and UI behaviors
• Focused on user-specific authorization and real-time integration with personal data

Worker Lifecycle Management

1. Creation

   • Universal Workers: Built through a structured builder UI with 5 configuration tabs
   • Specialized Workers: Built using the Canvas, via manual configuration or AI-assisted Builder Chat
   • Assistant Workers: Created by cloning a Universal Worker and adding personal account integrations

2. Deployment

   • Workers are deployed into the Worker List (Launchpad) for Users to interact with
   • Builders can set visibility (private/shared/public)
   • Specialized Workers can be embedded as Skills in other Workers

3. Versioning

   • Workers are versioned implicitly through configuration history and deployment
   • Future enhancements will include templating, rollback, and explicit version labels

4. Sharing

   • Universal Workers can be marked as public and shared across the organization
   • Specialized Workers are reusable as modules within other agents
   • Templates can be saved and reused for consistent setups across teams

Summary

The EverWorker architecture is designed for modularity, scalability, and flexibility. Universal Workers offer broad utility, Specialized Workers deliver precise automation, and Assistant Workers provide human-centric micro-assistance. Together, these components support a lifecycle that enables continuous creation, testing, reuse, and governance - all managed under a role-based platform that adapts to both technical and non-technical users.