RAG require a WPSOLR Enterprise license.

This documentation describe the configuration of your indexes for RAG.

WPSOLR RAG – Retrieval-Augmented Generation for WordPress & WooCommerce

Overview

WPSOLR RAG brings the power of Large Language Models (LLMs) directly into WordPress and WooCommerce websites.

It combines WPSOLR’s advanced search capabilities with Retrieval-Augmented Generation (RAG) to deliver AI-generated answers grounded in the actual content of your website.

Instead of relying exclusively on an LLM’s general knowledge, WPSOLR RAG retrieves relevant information from your website, provides it as context to an LLM, and generates contextual, relevant, and traceable answers.

This approach allows visitors to interact with your website through natural language, ask complex questions, and discover information more efficiently.

WPSOLR RAG turns your website into an AI-powered knowledge base.

Creating a WPSOLR RAG Gateway Connector

Configuration setting for LiteLLM

1. Install LiteLLM
2. Create a LiteLLM route (“wpsolr” here). You will select it in WPSOLR settings later.
   A route contains one or several LLM models from one or several providers.
   
3. Create a virtual key with access to your route (model”wpsolr” here). You will paste it in WPSOLR settings later.
   A virtual key secures access to the router, but also provides many other settings like usage statistics or budget caps.
   
4. Create a LiteLLM RAG connector in WPSOLR. Copy your virtual key and select your router created earlier:
   
5. (Optional) Prevent bots using your RAG with CloudFlare Turnstile or Google reCAPTCH 3:
   This is all you need to configure the LIteLLM connector. The remaining settings below are common to all parsing connectors, including LIteLLM and OpenRouter.

Configuration setting for OpenRouter

1. Create a an account at OpenRouter
2. Create your OpenRouter API key. You will paste it in WPSOLR settings later.
   
3. Create an OpenRouter Preset. You will select it in WPSOLR settings later.
   
   A Preset contains one or several LLM models from one or several providers. And many parameters like model temperature.
4. Create an OpenRouter connector in WPSOLR. Copy your API key and select your preset created earlier:
   
5. (Optional) Prevent bots using your RAG with CloudFlare Turnstile or Google reCAPTCH 3:
   

This is all you need to configure the OpenRouter connector. The remaining settings below are common to all parsing connectors, including LIteLLM and OpenRouter.

Activate the WPSOLR RAG add-on

The add-on will take care of formatting the search results sent to and the answer retrieved from the LLM.

Showing RAG Answers in results

RAG templates

RAG text templates

You can modify the RAG text templates in menu “2.6 Texts”. Those texts can also be translated with WPSOLR WPML or Polylang add-ons.

Citations Twig templates

For more advanced modifications, you can duplicate the WPSOLR Twig templates into your active theme’s directory.

RAG in search results

Select any search query already configured, from any search engine. With or without citations. Here we use an OpenSearch vector search configured with citations.

• Choose to insert the RAG answer in the search results automatically:
  or insert the RAG shortcode anywhere in the search results page:
  
• Choose your RAG Gateway connector on the  search connector:

• Choose a RAG template to display the AI-generated answer :
  
  This template renders as:
  

• Choose a RAG template for the waiting screen :
  
  This template renders as:
  
• Choose a RAG template for the prompt :
  
  or create your own prompt:
  
• Other parameters:
  

RAG in suggestions

• All suggestions created on a search query with a RAG will automatically show the RAG answers too
• The RAG answer will appear on the suggestions managed by this search connector:
  

Frequently Asked Questions