Use the Knowledge Base screen to give Dragwyb Click to Chat additional material for answering visitor questions. You can combine direct text, web page URLs, uploaded training documents, and selected public WordPress content, then build an index so the assistant can retrieve relevant passages.
Before you start
- Enable the AI Assistant module and open Click to Chat → AI Assistant → Knowledge Base.
- Configure an OpenAI or Google Gemini API key if you plan to use that provider for embeddings.
- Choose source material that is accurate, current, and appropriate for the chatbot to use.
Add knowledge sources
- Open the Knowledge Sources tab.
- Under Direct Text Knowledge, add factual content such as company details, product guidance, policies, or frequently asked questions.
- Under Web Pages (URLs), add the page URLs you want the plugin to process.
- Under Training Documents (Files), upload supported files such as PDFs, Word documents, CSVs, or text files.
- Open Database & Indexing and select the public website content types you want indexed. The list shows types with published items.
- Choose the embedding provider and vector storage option. The current interface offers OpenAI
text-embedding-3-smallor Google Geminigemini-embedding-001, plus SQLite or Pinecone storage. - Save the settings using the AI Assistant settings save action, then select Index Content Now.
Check indexing and test answers
When indexing is queued, the screen can display progress for document indexing and embedding generation. Wait for the job to finish, review the available indexing status, and test with a question whose answer is explicitly present in one of the sources. If the answer is missing or irrelevant, check that the correct source is included and that indexing completed successfully.
Changing the selected website content types may require re-indexing for the new selection to take effect. Automatic synchronization is controlled separately in Database & Indexing. See Refresh the Website Knowledge Base for update behavior and Improve Knowledge Retrieval Quality for troubleshooting advice.
For a complete overview of vector storage, retrieval thresholds, and source links, see Knowledge Base and Retrieval.