No code Knowledge Base

Lyzr Studio’s Knowledge Base feature lets you create, populate, and query a persistent repository of documents and data for Retrieval-Augmented Generation (RAG) and Semantic Modeling.

​

1. Create a Knowledge Base

First, define your new KB by providing basic metadata and selecting a vector store backend.

• Name: A unique identifier for your KB (e.g., “Product Docs”).
• Description: Brief summary of the KB’s contents or purpose.
• Vector Database: Choose where embeddings will be stored (e.g., Qdrant or Weaviate).

Click Create to provision your Knowledge Base.

​

2. Add Content & Parsing

Next, ingest data by uploading files or pointing to external sources. LyZR will auto-select the appropriate parser based on file type.

• Upload: Drag-and-drop or select file(s) (PDF, DOCX, TXT, CSV, JSON, etc.).
• Parser Selection: LyZR auto-detects format (e.g., text, table, JSON) and chooses the right parser.
• Indexing: Content is broken into chunks, embedded, and stored in the vector DB.

​

3. Knowledge Base Retrieval

Once populated, you can query the KB directly within Studio to retrieve relevant chunks.

• Search Bar: Enter a query and press Enter.
• Number of Chunks: How many results to return (default: 10).
• Retrieval Type: Choose between Basic (standard similarity search) or advanced modes.
• Score Threshold: Filter out low-relevance results.

By default, this setup creates a turnkey RAG pipeline. To customize retrieval strategies or integrate with your own agents, scroll down to the Custom RAG section.

​

4. Enable Semantic Data Model (Text-to-SQL)

You can also use the Knowledge Base as a Semantic Data Model to power Text-to-SQL agents that query structured data (like databases).

​

Enable Semantic Model Option

Toggle the “Use as Semantic Data Model” option in the KB setup screen.

This allows AI agents to perform structured queries using schema and context.

​

5. Connect a Database

To use Text-to-SQL features, link the KB to a live database.

• Navigate to Data Connectors under Studio.
• Click Create New if no databases are shown.
• Enter:
  • Database Type (PostgreSQL, MySQL, etc.)
  • Connection Credentials
  • Database Name

Once connected, the KB can read schema and data from your database.

​

6. Add Schema Documentation Agent

Schema understanding is enhanced by adding a Schema Documentation Agent.

• Click Create New under Schema Documentation Agent.
• Select your preferred LLM (e.g., GPT-4, Claude).
• The agent will generate column-level and table-level descriptions.

This step ensures your AI agent has context when writing SQL queries.

​

Summary

With these tools, you can build powerful no-code semantic retrieval and Text-to-SQL systems in Lyzr Studio — combining unstructured and structured data understanding with ease.