Overview

Trainkit is NUWA's AI knowledge management platform that enables robots to answer questions using RAG (Retrieval-Augmented Generation).

By uploading product information, FAQs, service documentation, or any enterprise knowledge into Trainkit, you can create a dedicated GenAI project. Combined with the Context-Aware Dialogue module in Roflow, the robot can understand users' questions in natural language, retrieve relevant information from the knowledge base, and generate appropriate responses.

In this architecture:

• Trainkit — Defines the AI's personality and manages the knowledge base.
• NUWA GenAI — Understands user questions, retrieves relevant information from the knowledge base, and generates responses.
• Roflow — Designs the robot's interaction flow and integrates conversations between users and NUWA GenAI.

The complete workflow is:

User asks a question → Roflow receives the request → NUWA GenAI understands the question → Relevant knowledge is retrieved from Trainkit → A response is generated → The robot speaks the response

After completing the setup, you can build knowledge-based conversational applications for product introductions, customer support, guided tours, education, employee training, and internal knowledge assistance.

Steps

A. Create a GenAI Project in Trainkit

1. Sign in to Trainkit.

2. Create a new GenAI Dialogue project.

3. Select Personality Settings from the left sidebar to configure the AI's role and personality.

   Describe the AI's role, target audience, and response guidelines according to your application. For example:

   <aside>

   You are a product support assistant responsible for answering questions about product features, specifications, and usage. Always prioritize information from the knowledge base and respond in a concise, friendly, and easy-to-understand manner.

   </aside>

   

B. Build a Knowledge Base

1. Select Knowledge Base from the left sidebar.

2. Upload the knowledge documents you want the AI to reference.

   Examples include:

   • Product introductions and specifications
   • User manuals
   • FAQs
   • Service information
   • Company or brand information
   • Educational and training materials
   • Any other domain-specific knowledge

3. After uploading your knowledge, click Train.

4. It is recommended to validate the AI before integrating it with Roflow.

   Select Training Results from the left sidebar and verify that the AI can correctly answer questions using your knowledge base.

   For example:

   <aside>

   How long can this robot operate on a single charge?

   </aside>