RNet AI
  • INTRODUCTION
    • About Rnet
    • Quickstart
    • Core Concepts
      • List of Model Providers
  • GETTING STARTED
    • Model
      • Add New Provider
      • Predefined Model Integration
      • Custom Model Integration
      • Interfaces
      • Schema
    • Application Orchestration
      • Overview
      • Interactive Chat Application
      • Agent
      • App Kits
    • Workflow
      • Core Concepts
      • Node Overview
        • Start
        • Question Classifier
        • Knowledge Retrieval
        • Variable Aggregator
        • LLM
        • Direct Reply
        • IF/ELSE
        • HTTP Request
        • End
    • Knowledge Base
Powered by GitBook
On this page
  1. GETTING STARTED
  2. Application Orchestration

Agent

PreviousInteractive Chat ApplicationNextApp Kits

Last updated 8 months ago

An Agent Assistant can leverage the advanced reasoning abilities of large language models (LLMs) to perform tasks autonomously. Think about an AI Agent like a Smart robot helper. This robot doesn't just do what you tell it—it can also figure out what needs to be done, break down big jobs into smaller tasks, and use different tools to get things done. Plus, it gets better and smarter the more it works with you.

To facilitate quick learning and use, application templates for the Agent Assistant are available in the ‘Explore’ section. You can integrate these templates into your workspace. The new Rnet ‘Studio’ also allows the creation of a custom Agent Assistant to suit individual requirements. This assistant can assist in analysing financial reports, composing reports, designing logos, and organising travel plans.

Orchestrating Prompts for the Agent Assistant: In the “Instructions” section, you can define the task objectives, workflow, resources, and limitations for the Agent Assistant. This ensures optimal results by providing clear guidelines for the agent’s actions.

Adding Tools for the Agent Assistant: In the “Context” section, you can integrate knowledge base tools that aid in information retrieval, enhancing the agent’s background knowledge. The “Tools” section allows for the addition of tools that expand the capabilities of Language Learning Models (LLMs), such as internet searches, scientific computations, or image creation. Rnet supports both built-in and custom tools including custom API tools compatible with OpenAPI/Swagger and OpenAI Plugin standards.

Tools enable the creation of more robust AI applications on Rnet. They allow the Agent Assistant to perform complex tasks through reasoning, step decomposition, and tool invocation. Additionally, these tools facilitate the integration of applications with external systems or services, allowing for activities like code execution and access to exclusive information sources.

Agent Settings: Rnet provides two inference modes for the Agent Assistant: Function Calling and ReAct. Models supporting Function Calling, such as GPT-3.5 and GPT-4, offer better and more stable performance. For models not supporting Function Calling, the ReAct inference framework is used to achieve similar results. The Agent settings also allow for modification of the agent’s iteration limit.

Configuring the Conversation Opener: You can set up a conversation opener and initial questions for your Agent Assistant. This feature displays at the start of a user’s first interaction, highlighting the types of tasks the agent can perform and providing example questions.

Debugging and Preview: Before publishing the Agent Assistant as an application, you can debug and preview it. This step allows you to test the agent’s effectiveness in completing tasks and make necessary adjustments.

Application Publish: After debugging your application, click the “Publish” button in the top right corner to create a standalone AI application. In addition to experiencing the application via a public URL, you can also perform secondary development based on APIs, embed it into websites, and more.

Adding Tools for the Assistant