Skip to main content

Getting Started

GenAI RAG is a conversational AI that searches across your Nexla nexsets, reasons over the results, and generates answers with inline citations. You can use it through the web UI at genai.nexla.com or the API.

Prerequisites​

Before you start, you need:

  • A Nexla account with at least one active nexset (vector, SQL, or REST API source)
  • A service key for authentication — create one at dataops.nexla.io/settings/authentication
  • At least one LLM credential configured in your Nexla account (OpenAI, Anthropic, Google, Azure, or Mistral)
Protect Service Keys

Service keys are equivalent to your account password. Store them securely and treat them as highly sensitive information.

1. Authenticate​

  1. Navigate to genai.nexla.com.
  2. Click Connect to Get Started or open Settings (gear icon) and go to the Authentication tab.
  3. Paste your Nexla service key into the Service Key field.
  4. Click Connect.

Once authenticated, you see a green Authenticated badge and the sidebar becomes available.

2. Select an LLM Credential​

  1. In the sidebar, locate the Provider section.
  2. Click a credential card (if you have one or two) or use the dropdown (if you have more than two).
  3. The model selector auto-populates with the top-ranked model for that provider. To change it, use the model dropdown.

For details on available models and trade-offs, see Available Models.

3. Add Data Sources​

  1. In the sidebar, locate the Data section.
  2. Start typing a nexset name or ID in the search field.
  3. Click a nexset in the dropdown to add it. Selected nexsets appear as removable chips.
  4. Repeat to add additional nexsets. You can query one or more nexsets in a single session.
Supported Data Sources

GenAI RAG works with vector databases, SQL databases (Query Mode), and REST API nexsets. For SQL sources, the query must follow the format {query = <SQL query>}.

4. Ask a Question​

  1. Type your question in the chat input at the bottom of the window.
  2. Press Enter to send (or Shift+Enter for a new line).
  3. The agent thinks, searches your nexsets, and streams a response with inline citation markers like [1].

You can click any citation badge to see the source data, or continue the conversation with follow-up questions.

Next Steps​

  • Chat & Querying — full guide to the chat interface, streaming, citations, and message actions
  • Canvas Panel — exploring citations, sources, and tool call details
  • Settings — authentication, response tuning, and filter management
  • Agentic RAG API — programmatic access via REST API