Skip to main content

step-by-step-guide

sidebar_position: 2

Step- by- Step: Set Up Knowledge Base & RAG

This comprehensive guide walks you through setting up a Knowledge Base with RAG (Retrieval Augmented Generation) in Axie Studio.

Overview​

Time Required: 10- 15 minutes What You'll Create: A knowledge base that enables AI to answer questions using your documents Result: AI- powered Q&A with source citations

Step 1: Navigate to Knowledge Base​

What You'll See​

  1. Sign in to your Axie Studio account
  2. In the dashboard sidebar, click "Knowledge Base" (or "Agent Configuration")
  3. You'll see the Knowledge Base page with:
  • A title bar: "AGENT CONFIGURATION"
  • Two- column layout:
  • Left: Configuration options
  • Right: Chat preview

Visual Description​


 AGENT CONFIGURATION 
 Configure your AI agent's capabilities and connections 

 
 
 Configuration Chat Preview 
 (Left Column) (Right Column) 
 
 [Connection Banner] [Chat Interface] 
 [Config Sections] [Messages] 
 [Input Field]

Step 2: Connect a Chat Interface​

What You'll See​

If no chat interface is connected, you'll see a Connection Banner:

Visual:


 No Chat Interface Connected 
 
 You need to connect a chat interface to configure 
 your knowledge base. 
 
 [Connect Interface] button

Step 2.1: Click "Connect Interface"​

Button: "Connect Interface" Location: In the connection banner

What Happens:

  • Opens a dialog
  • Shows list of your chat interfaces
  • Allows you to select one

Step 2.2: Select Chat Interface​

Dialog: "Connect Chat Interface"

What You'll See:

  • List of your chat interfaces
  • Each interface shows:
  • Name
  • Slug
  • Status (Active/Inactive)

Visual:


 Connect Chat Interface 
 
 
 Select a chat interface to connect: 
 
 Customer Support 
 slug: customer- support 
 Status: Active 
 
 Sales Chat 
 slug: sales- chat 
 Status: Active 
 
 [Cancel] [Connect]

Step 2.3: Confirm Connection​

Button: "Connect" What Happens:

  • Connects the interface
  • Creates knowledge base configuration (if needed)
  • Banner disappears
  • Configuration sections appear

Step 3: Configure Core Settings​

What You'll See​

After connecting, you'll see Configuration Sections organized in an accordion:

  1. Tool Overview
  2. Core Settings (System Prompt, LLM)
  3. Advanced Settings (Embeddings, Vector Store, Parser)
  4. Helper Tools (Webhook, Chat Input/Output, etc.)
  5. Connections (Composio, Form Builder)

Step 3.1: Configure System Prompt​

Section: "Core Settings" "System Prompt"

What You'll See:

  • A card showing current system prompt
  • "Configure" button

Visual:


 System Prompt 
 
 You are a helpful AI assistant. 
 
 [Configure] button

Action:

  1. Click "Configure"
  2. Dialog opens with text area
  3. Enter your system prompt
  4. Click "Save"

Example System Prompt:

You are a helpful customer support assistant for Acme Corporation. 
Answer questions using the provided documents. Always cite your sources.
Be friendly, professional, and concise.

Step 3.2: Configure LLM Provider​

Section: "Core Settings" "LLM Provider"

What You'll See:

  • Card showing current LLM settings
  • Provider dropdown
  • Model selection
  • API key field

Visual:


 LLM Provider 
 
 Provider: [OpenAI ] 
 Model: [GPT- 4 ] 
 API Key: [] [Configure] 
 
 [Configure] button

Action:

  1. Click "Configure"
  2. Dialog opens with:
  • Provider dropdown (OpenAI, Cloudflare Workers AI, DeepSeek)
  • Model dropdown (depends on provider)
  • API Key field
  1. Select provider and model
  2. Enter API key
  3. Click "Save"

Recommended:

  • Provider: OpenAI
  • Model: GPT- 4 (or GPT- 3.5- turbo for cost savings)
  • API Key: Your OpenAI API key

Step 4: Configure Advanced Settings​

Knowledge Base Complete View

The Knowledge Base page shows your complete RAG configuration. The left column contains all configuration sections, the right column shows a live chat preview, and the bottom section displays a visual workflow builder that illustrates how your RAG pipeline components are connected. The Advanced Settings section contains configuration options for embeddings, vector store, document parser, and other advanced RAG components.

Step 4.1: Configure Embeddings​

Section: "Advanced Settings" "Embeddings"

What You'll See:

  • Card showing embedding model
  • API key field

Visual:


 Embeddings 
 
 Model: voyage- large- 2 
 API Key: [] [Configure] 
 
 [Configure] button

Action:

  1. Click "Configure"
  2. Dialog opens with:
  • Model dropdown (voyage- large- 2, voyage- 3- large, etc.)
  • API Key field
  1. Select model
  2. Enter Voyage AI API key
  3. Click "Save"

Step 4.2: Configure Vector Store​

Section: "Advanced Settings" "Vector Store"

What You'll See:

  • Card showing vector store settings
  • ChromaDB configuration

Visual:


 Vector Store 
 
 Provider: ChromaDB 
 Collection: [your- collection- name] 
 API Key: [] [Configure] 
 
 [Configure] button

Action:

  1. Click "Configure"
  2. Dialog opens with:
  • API Key field
  • Tenant field
  • Database field
  • Collection name field
  1. Enter ChromaDB credentials
  2. Click "Save"

Step 5: Upload Documents​

What You'll See​

In the right column, you'll see the Chat Preview with an upload section.

Step 5.1: Access Upload Area​

Location: Right column, below chat preview

What You'll See:

  • File upload area
  • Drag- and- drop zone
  • "Upload Documents" button

Visual:


 Upload Documents 
 
 
 
 
 Drag files here or click to upload 
 
 Supported: PDF, DOCX, TXT 
 Max size: 10MB per file 
 
 
 
 [Upload Documents] button

Step 5.2: Upload Files​

Action:

  1. Click "Upload Documents" or drag files
  2. File picker opens
  3. Select files (PDF, DOCX, or TXT)
  4. Files upload automatically

What Happens:

  • Files are processed
  • Text is extracted
  • Documents are chunked
  • Embeddings are generated
  • Documents are stored in ChromaDB

Progress Indicator:

Uploading: document.pdf
Processing... [] 80%

Step 5.3: Verify Upload​

What You'll See:

  • Success message
  • Document appears in list
  • Status: "Processed" or "Processing"

Visual:


 Uploaded Documents 
 
 
 document.pdf 
 Status: Processed 
 Chunks: 45 
 
 large- document.pdf 
 Status: Processing... 
 Progress: 60%

Step 6: Test Your Knowledge Base​

What You'll See​

In the Chat Preview (right column), you can test your knowledge base.

Step 6.1: Ask a Question​

Location: Chat input field in preview

Action:

  1. Type a question related to your documents
  2. Press Enter or click Send

Example:

What is your return policy?

Step 6.2: See AI Response​

What You'll See:

  • AI processes the question
  • Searches your documents
  • Generates answer
  • Shows source citations

Visual:


 User: What is your return policy? 
 
 AI: Our return policy allows returns within 30 days 
 of purchase. You can request a refund by 
 contacting support@yourcompany.com. 
 
 Source: Returns Policy.pdf, page 3

Step 7: Enable Tools (Optional)​

What You'll See​

Section: "Tool Overview"

Visual:


 Tool Overview 
 
 
 Core Tools 
 search_knowledge_base 
 confirmation_required 
 
 Advanced Tools 
 web_search 
 code_execution 
 
 [Manage Tools] button

Step 7.1: Open Tool Overview​

Button: "Manage Tools" Action: Click to open tool management dialog

![Manage Tools Dialog](/img/screenshots/manage- tools- dialog- detailed.png)

Step 7.2: Enable Tools​

Dialog: "Tool Overview"

What You'll See:

  • List of all available tools
  • Toggle switches for each tool
  • Tool categories

Action:

  1. Toggle tools on/off
  2. Click "Save"

Visual Workflow Summary​


 STEP 1: Navigate to Knowledge Base 
 Dashboard Knowledge Base 

 

 STEP 2: Connect Chat Interface 
 [Connect Interface] Select interface 

 

 STEP 3: Configure Core Settings 
 - System Prompt 
 - LLM Provider & Model 
 - API Keys 

 

 STEP 4: Configure Advanced Settings 
 - Embeddings (Voyage AI) 
 - Vector Store (ChromaDB) 
 - Parser settings 

 

 STEP 5: Upload Documents 
 Drag & drop or click to upload 
 Wait for processing 

 

 STEP 6: Test Knowledge Base 
 Ask questions in chat preview 
 Verify answers and sources 

 

 STEP 7: Enable Tools (Optional) 
 Manage tools in Tool Overview 
 Enable/disable as needed

Common Questions​

How long does document processing take?​

Usually 1- 2 minutes per document, depending on size.

Can I upload multiple documents?​

Yes! Upload as many as you need. Each is processed separately.

What file formats are supported?​

PDF, DOCX, and TXT files. Max 10MB per file.

How do I know if documents are processed?​

Check the status indicator. "Processed" means ready to use.

Can I delete documents?​

Yes, from the document list. Click the delete icon.

Next Steps​

Ready to set up your knowledge base? Go to Dashboard Knowledge Base and get started!