Create Vectors
Vectors in Indite empower your AI chatbots with custom knowledge bases for precise, context-aware responses. Follow this guide to set up your first vector and enhance your bot’s capabilities with retrieval-augmented generation (RAG).
Steps to Create a Vector
Access Vector Configuration
- Log in to your Indite dashboard.
- Navigate to the Vectors section in the sidebar.
- Click New Vector to start.
Set Vector Name
Enter a unique name, such as Product_Docs_Vector, to identify your knowledge base.
Add Text Content
Input long-form text like product descriptions, FAQs, or support articles.
Text content is automatically split into chunks for efficient embedding and retrieval.
Add Q&A Pairs
Upload Excel or CSV files containing targeted question-answer pairs for specific knowledge.
Configure Website Crawler
Enter a website URL and set the maximum crawl depth to extract content automatically. Use Individual Links to include specific pages.
Upload Files
Add PDFs, Excel, Word, or CSV files to enrich your vector’s knowledge base.
Set Advanced Parameters
Configure processing settings:
- Chunking Method: Recursive Text Splitter
- Chunk Size: 1000 characters (default)
- Overlap: 100 characters (default)
Train Vector
Click Train Vector to process your content and generate embeddings for use in your chatbot.
Demo Example
- Vector Name: Product FAQ
- Text Content: Introduction to company products
- Q&A: 3 sample question-answer pairs
- Crawl: Company website, depth 1
- Upload: Product manual PDF
- Train: Vector ready for chatbot integration
Ensure your data sources are clean and relevant to optimize vector performance and chatbot accuracy.