Agent: An Intelligent Workflow That Understands, Decides, and Acts

An Agent is an intelligent partner built on Large Language Models (LLMs) that understands user goals, makes autonomous decisions, and executes real tasks.

While an Ability performs defined actions according to a user-specified flow, an Agent determines and executes the most effective actions on its own.

Agents can maintain conversational context using memory, and testing can be conducted through a chat interface, allowing for natural, human-like interaction.

Through Agents, users can build automated task systems, and completed agents can be deployed via API to power various services.

Key Features and Advantages

• Chat Interface : Enables interactive testing and immediate feedback through conversations.

• Conversation generation trace: Provides a detailed view of how the agent generates conversations, enabling a clear understanding of the specific logic.

• Intelligent Decision-Making : Analyzes situations using LLMs to plan and select the optimal course of action.

• Memory Utilization : Stores information in short-term and long-term memory to maintain context from previous conversations or tasks.

• Scalability : Combines with various nodes to create and expand complex automation scenarios.

Example Use Cases

• Personal Assistant : Remembers conversations and provides personalized responses.

• Workflow Automation : Performs repetitive tasks such as data analysis, report generation, and email handling.

• Enterprise Solutions : Builds intelligent systems for customer support or decision-making assistance.

Learn More

The Advanced Guide is composed of agent-focused tutorials.

If you want to build a high-level automation system using agents, refer to the agent tutorials, which are structured around practical, real-world examples.