Agent Overview

Agents are the core of AnchorAutomate - intelligent AI entities that can execute tasks and workflows on your behalf. Each agent can be configured with specific instructions, connected to various applications, and assigned different task definitions.

What is an Agent?

An agent is an AI-powered assistant that:

• Understands Instructions: Agents can interpret natural language instructions and execute complex workflows
• Connects to Apps: Agents can interact with external applications through the App Marketplace
• Requests Help: When uncertain, agents can ask for human assistance or approval
• Learns from Feedback: Agents can be fine-tuned based on user feedback and corrections

Agent Types

Cloud Agents (Default)

Cloud agents run on AnchorAutomate's infrastructure and are available to all users:

• No Setup Required: Start using immediately after creation
• Automatic Updates: Always running the latest AI models and features
• Scalable: Handle multiple tasks simultaneously
• Secure: Enterprise-grade security and data protection

On-Premise Agents (Enterprise Only)

Enterprise customers can deploy agents on their own infrastructure:

• Enhanced Security: Keep sensitive data within your network
• Custom Configuration: Configure agents for specific infrastructure needs
• Local LLM Integration: Connect to your own language models
• STDIO MCP Support: Integrate with local services and tools

Cloud vs On-Prem Agents

Cloud Agents

• Hosted and managed by Anchor Automate.
• Require no setup—just assign and run tasks.
• Suitable for most organizations unless specific compliance or data residency rules apply.

On-Prem Agents

• Deployed inside your own infrastructure.
• Ideal for sensitive environments with strict compliance or internal data policies.
• Still connect to Anchor Automate for task scheduling and reporting.

Choose the option that best fits your organization's infrastructure, data policy, and security needs.

Enterprise Feature

On-premise agents are only available on the Enterprise Plan. Contact your account manager for more information about Enterprise features.

Agent Capabilities

Task Execution

• Execute predefined task definitions
• Handle dynamic inputs and parameters
• Manage complex multi-step workflows
• Provide status updates and progress reports

App Integration

• Read and write data across connected applications
• Perform actions like sending emails, creating documents, updating records
• Handle authentication and API interactions automatically

Human Interaction

• Request clarification when instructions are unclear
• Ask for approval before executing sensitive operations
• Provide detailed explanations of actions taken
• Accept feedback and corrections

Best Practices

Agent Design

1. Specific Role: Organize agents into specific roles (similar to human roles)
2. Appropriate Scope: Don't make agents too broad or too narrow
3. Security First: Use human-in-the-loop for sensitive operations

App Connections

1. Minimal Permissions: Only connect apps that are necessary
2. Separate Instances: Use different app instances for different purposes
3. Regular Audits: Review connected apps periodically
4. Secure Credentials: Use proper authentication methods

Next Steps

• Create your first agent
• Connect Apps to Your Agent
• Define your first task
• Learn about On-Premise deployment (Enterprise only)
• Explore the App Marketplace