Data Source Node

The Data Source Node plays the role of fetching external data. It can perform API calls, database queries, file reading, etc.

📋 Data Source Node List

VectorDB

Add data for AI to reference

✨ Key Features

• ✅ Upload documents and save to Vector DB
• ✅ Auto-search documents related to the question
• ✅ Prompt Auto-Completion (Optimized for RAG)

🔧 Configuration

📋 Supported File Formats

• ✅ PDF
• ✅ TXT
• ✅ Markdown (.md)

🎯 Usage

1. Upload Documents

   • Click VectorDB node
   • "Add File" button
   • Upload documents (Max 10MB)

2. Auto Vectorization

   • Uploaded docs are automatically vectorized
   • Semantic search enabled

3. Search & Prompt Generation

   • Searches relevant docs based on the previous node's input (question).
   • Important: Combines searched content and user question to generate a complete prompt text.

📝 Result Data Structure

This node returns not just simple document content, but a formatted prompt text ready to send to LLM ({{result}}).

Components:

1. Searched related document chunks ([context])
2. User question ([query])

Example ({{result}} value):

[context]: 
Similarity: 0.8123
(Reference Doc Content 1...)
==========
Similarity: 0.7654
(Reference Doc Content 2...)

---

[query]: User entered question

💡 Usage Example

Document-based Q&A Chatbot

[Text Input] → [VectorDB] → [ChatGPT] → [Chatbot Response]

1. User enters question
2. VectorDB finds related docs and forms prompt
3. ChatGPT receives `` and generates answer without extra prompting

ChatGPT Config Example:

prompt: {{result}}
(system prompt only needs role definition)

Http Request

Sends an HTTP request including JSON Body

✨ Key Features

• ✅ Call REST API
• ✅ Support GET, POST, PUT, DELETE
• ✅ Custom Header settings
• ✅ Send JSON Body

🔧 Required Settings

📡 GET Request Example

Call Weather API

URL: https://api.openweathermap.org/data/2.5/weather?q=Seoul&appid=YOUR_API_KEY
Method: GET
Headers: {'Content-Type': 'application/json'}
Body: (Empty)

📡 POST Request Example

Call External AI API

URL: https://api.example.com/generate
Method: POST
Headers: {
  'Content-Type': 'application/json',
  'Authorization': 'Bearer YOUR_TOKEN'
}
Body: {
  "prompt": "{{result}}",
  "max_tokens": 100
}

📝 Result Data

Access API response via {{result}} in the next node

Example: {{result}} → API response text (JSON String, etc.)

💡 Usage Example 1: Weather-based Recommendation

[Click] → [Http Request] → [ChatGPT] → [Telegram]

1. Call Weather API via Http Request
2. ChatGPT recommends outfit based on weather
3. Send to Telegram

💡 Usage Example 2: News Summary

[Click] → [Http Request] → [ChatGPT] → [Slack]

1. Fetch latest news from News API
2. ChatGPT summarizes in 3 lines
3. Send to Slack

🎯 Pro Tips

API Key Security Management

• ❌ Exposed in URL: ?api_key=123456
• ✅ Send via Header: {'Authorization': 'Bearer 123456'}

Dynamic URL Generation

https://api.example.com/user/{{result}}

🆚 VectorDB vs Http Request

🔗 Connecting with Other Nodes

VectorDB + AI Generation

[Question] → [VectorDB] → [ChatGPT] → [Response]

Since VectorDB already creates the prompt, use {{result}} as is in the AI node.

Http Request + AI Generation

[Start] → [Http Request] → [Claude] → [Slack]

AI analyzes the data received from the API.

🎯 Next Steps

• Text Generation Node - Generate text with AI
• Image Generation Node - Generate images with AI
• Output Node - Send results

❓ FAQ

Q: How many files can I upload to VectorDB? A: Currently, 10MB per file, up to 100MB total.

Q: What happens if Http Request times out? A: It automatically fails after 30 seconds, and the workflow stops.