Skip to main content

Create a Vector Database

Instructions​

Step 1: Access the Vector Database Section​

  1. Log in to your VAKStudio account.
  2. Navigate to the Vector Database section from the main navigation menu.
  3. Click on the New Embeddings button to start creating a new Vector Database.

Step 2: Configure Your Vector Database​

  1. Name: Enter a unique name for your Vector Database.
  2. Description: Provide a brief description of your Vector Database.
  3. Type: Select the type of data:
    • CSV
    • Markdown
  4. Select Folder:
    • Based on the type selected, you'll see a list of folders from the Datastore containing the corresponding file types.
    • Choose the folder you want to use for creating embeddings.

Step 3: Additional Options for Markdown Files​

If you select Markdown as the type:

  • Heading Levels: You'll have checkboxes for headings H1 to H6.
    • Purpose: Choose where to split the text.
    • Example: Selecting H2 and H3 will split the text at every H2 and H3 heading, creating embeddings for each section.

When uploading markdown files to create a Vector Database, it's essential to organize the content efficiently for embedding generation. The VAKStudio system splits markdown text at logical points using headings (H1-H6) to create smaller, meaningful chunks of content. By structuring your markdown files properly, you can reduce token usage, speed up embedding generation, and improve the accuracy of your AI's responses.

Using the Vector Database​

Integration with InputPrompt Node​

The embeddings stored in the Vector Database can be used as context in the InputPrompt node within your AI workflows in VAKFlows. This significantly enhances AI-generated responses by providing relevant data and context.

Key Use Cases:​

  • Contextual AI Responses: Improve the accuracy and relevance of AI-generated responses.
  • Efficient Data Processing: Use embeddings to process large datasets with more precise, context-aware outcomes.