Agent Overview

In the GPTBots platform, an Agent is a native AI application designed for human users, equipped with capabilities such as natural language understanding, multi-turn interaction, short and long-term memory, tool invocation, task planning, quality inspection, and reflection. Developers can quickly build efficient and intelligent enterprise AI applications, significantly enhancing enterprise work efficiency and intelligence levels.
GPTBots provides three types of Agents: Agent, FlowAgent, and MultiAgent, to meet the AI application needs of enterprises in business scenarios with varying complexities. Developers can flexibly choose Agent, FlowAgent, or MultiAgent based on actual business goals to quickly build efficient and intelligent enterprise-grade AI applications.

graph TD
    A[Agent] -->|Single LLM| B[Regular Conversations]
    C[Flow-Agent] -->|Multi-component/Multi-LLM| D[Complex Business Process Orchestration]
    E[Multi-Agent] -->|Multi-role Collaboration| F[Autonomous Work of Intelligent Teams]

Agent

Agent is the most basic type of Agent in the GPTBots platform. It drives business processes through a single LLM, supporting natural language input, multi-turn conversations, context, memory, and tool invocation. Suitable for simple business scenarios in enterprises, the features and application scenarios of Agent include:
Features of Agent:

• Understands and responds to users' natural language queries
• Conducts multi-turn interactions, maintaining conversation context
• Invokes knowledge bases, external tools, or APIs to provide professional answers

Typical Application Scenarios:

• Intelligent customer service
• FAQ Q&A
• Enterprise knowledge retrieval

FlowAgent

Flow-Agent is a type of Agent in the GPTBots platform specifically designed for complex dialogue processes and multi-turn interaction scenarios. It combines the contextual understanding capabilities of conversational AI with visual process orchestration, suitable for intelligent applications requiring complex business logic handling and multi-step task execution.

Features of FlowAgent:

• Supports process orchestration, ideal for scenarios involving multi-step, conditional branching, and complex logic
• Componentized design, with each component encapsulating specific business capabilities (e.g., knowledge retrieval, logic judgment, preset replies, manual transfer, etc.)
• Supports multi-LLM collaboration, enhancing the professionalism and accuracy of responses

Typical Application Scenarios:

• Intelligent customer service requiring multi-turn dialogue and complex business logic
• Form bots for automated collection and processing of user information
• Enterprise assistants for multi-step task execution and decision-making
• Enterprise complex business process automation

MultiAgent

Multi-Agent is an enterprise-grade multi-role AI Agent system launched by the GPTBots platform, designed for automation and intelligent collaboration in complex business scenarios. Multi-Agent simulates an "AI team" to achieve autonomous perception, task planning, division of labor, execution, and reflection, providing enterprises with flexible and efficient AI solutions.

Note: MultiAgent is currently in beta testing. Stay tuned.

Features of MultiAgent:

• Supports customizing multiple specialized AI roles (e.g., development, product, testing, data analysis, etc.) for each business goal, with roles collaborating and synchronizing information to complete complex tasks.
• Can automatically decompose tasks, assign roles, track progress, and integrate results based on set task goals, achieving end-to-end automation from goal setting to result output.
• Equipped with environmental awareness and contextual memory capabilities, dynamically adjusting strategies during task execution to improve task accuracy and intelligence.
• Comes with various pre-configured Agent roles (e.g., AI Coder IDE, Browser Use, Computer Use, etc.), ready to use out of the box.

Typical Application Scenarios:

• In-depth research and report generation
• Data insights and analysis
• Automated generation of documents such as bids and financial details
• AI research assistants for roles like product managers and marketing directors