Transform Your Pinecone Vector Database RAG Model into a Live WordPress AI Chatbot!

Want to showcase your custom Pinecone RAG model to the world? This WordPress AI plugin lets you instantly deploy a personalized chatbot that leverages your Pinecone vector database to deliver precise, context-aware responses.

Create a custom ChatGPT-powered AI chatbot for your WordPress website in minutes! Simply add the WordPress chatbot block, connect your Pinecone database and OpenAI LLM, and you're ready to engage visitors with your specialized knowledge base.

See the plugin in use as a project management tool chatbot here.

Key Features of this WordPress AI Chat Plugin:

• Cost-Efficient AI for WordPress - Intelligent filtering ensures OpenAI calls only happen for relevant queries that match your Pinecone RAG model's knowledge base, preventing unnecessary API costs.
• Quality-Focused Custom ChatGPT Plugin - You control the semantic match threshold between user questions and your vector database responses, ensuring your chatbot for business delivers only high-quality, relevant answers.
• Versatile Custom AI Chatbot Solution - Deploy multiple WordPress AI chatbots across different pages, each connected to distinct Pinecone vector databases and OpenAI models for specialized topic coverage.
• Extensively Customizable - Fine-tune your custom chatbot with 23 configuration options spanning display, Pinecone embedding, vector database querying, and OpenAI parameters.
• Rapid WordPress AI Plugin Integration - Get your personalized chatbot up and running in minutes with our streamlined installation process.
• Flexible Access Control - Choose between public-facing chatbots to share your knowledge or private deployments for internal use.

Process

The Custom AI Chatbot plugin is more than just a prompt field.  Prompts go through a six-step process with three API calls.

• Embed User Prompt - The user prompt is embedded into a vector using the Pinecone model of your choice.
• Retrieve Data w/Score - The prompt is matched to your custom Pinecone RAG model data.  Multiple records are returned with a quality score.
• Assess Data Quality - Records are filtered based on the quality score threshold you set.
• Don't Answer - Low quality responses are not passed to OpenAI.  The user receives a message that the prompt cannot be answered.
• Submit to OpenAI - High quality responses are appended to a system prompt (that you define), and are submitted to the OpenAI model of your choice.
• Answer Question - OpenAI responds based only on the records submitted.

Requirements

• Vector database on Pinecone
• Pinecone API key
• OpenAI API key
• WordPress site

Support

• This version of the plugin is not supported.
• You may submit support requests here, but support will only be provided at my discretion.
• Enhancement requests or other feedback are highly encouraged.
• If you are interested in a subscription version with full support, please contact me.