Quickstart

Step 1: Access Vector Management

1. Log into your Diaflow workspace

2. Navigate to the Vectors page on the sidebar

3. You’ll see two main tabs:

• Group: Default landing view showing organized collections

• Root: Shows all individual files and content

Vector Management Interface Alt text: Vector Management interface showing Group and Root tabs with sample vector files

Expected Result: You should see the Vector Management interface with the Group tab active by default.

Step 2: Add Your First Document

1. Click Create a vector group button

2. Create your group with Name and Description (if needed)

1. In your group which you just created, click the Add file button

1. Choose file type: Documents, URLs, or Articles

2. Upload a PDF, DOCX, TXT, MD, CSV, XLSX, XLS, or JSON file (under 50MB), crawl an individual URL/website sitemap, or write an article

3. For articles or URLs, you can configure chunking settings (optional):

• Chunking Method: Word (default), Letter, Sentence, or Passage

• Chunk Size: 200 (default, range: 0-4500)

• Chunk Overlap: 3 (default, range: 0-4500)

1. Click Create

2. Select Root tab to verify the file exists there after adding to Group

Document Upload Interface Alt text: Document upload modal showing file selection, chunking settings, and group assignment options

Best Practice: Manage information in specific groups for stable search performance, but files in Root allow searching across “All data from vectors” when you don’t remember which group contains specific knowledge.

Expected Result: Your document appears in the selected location with “Processing” status, then “Completed” when ready, or “Rejected” if over limitations.

Step 3: Test Vector Search

1. Navigate to the Flows tab at the sidebar, create a New Flow (App or Automation based on your objective)

2. Add a Diaflow Vector node

3. Configure:

• Action: Search data from Vectors

• From: All data from vectors, From vector groups (select data from specific groups), or From vector files (select data from specific files)

• Input: Enter a search/question query related to your uploaded document

1. Add an LLM node beside Diaflow Vector to receive filtered data and generate rich content

2. Add the output node

3. Execute the flow

Vector Search Flow Alt text: Flow builder showing Diaflow Vector node connected to LLM node with configuration panel open

Expected Result: The node returns relevant content from your uploaded document based on your search query and can analyze data based on the knowledge you provide at Vector with an LLM node.

Step 4: Integrate with AI Chat

1. Back to Flows page, Create New Flow with Chatbots type

1. You can start with Simple Mode, in the Knowledge Source section, select your vector sources

1. Choose scope: Root (all data from vector), Group (from vector groups), or specific Files (from vector files)

2. Check the LLM additional node to analyze information from the content you uploaded at the vector in the AI Engine and choose the model you want to use

1. Test the chatbot with questions related to your content in the chat field

AI Chat Configuration Alt text: Chat Simple Mode configuration showing Knowledge Source selection with vector groups

Expected Result: Your chatbot now responds using knowledge from your uploaded documents.