Search Knowledge Base by Keyword
This KB documentation is using Advanced Search, Links Editor and Elegant Layouts add-ons.
AI Training Data – Teach It Your Business
Overview
The AI Chat and AI Search are designed to assist your users, but to truly understand your business or organization, it needs to be trained.
This training process involves feeding the AI relevant materials such as:
- KB Articles
- Posts
- Pages
- PDFs
- FAQs
- Notes
By doing so, the AI can learn the nuances of your workflows, terminology, and customer needs, enabling it to provide accurate and tailored responses.
Think of it like onboarding a new team member—without the right information, the AI can only offer generic answers. Once trained with your materials, it becomes a knowledgeable partner, capable of addressing specific inquiries and supporting your team or customers more efficiently.
Organizing a Knowledge Base and AI Vector Store
Understanding Your Knowledge Sources
Every organization accumulates knowledge in diverse formats – from internal documents (PDFs, XML/CSV data, spreadsheets, code repositories) to support emails/notes and existing help-center articles:
Website Knowledge Base (KB): Existing help articles, FAQs, and guides on your WordPress site (the user-facing content).
Support and Reference Materials: Support tickets or emails, technical notes, product manuals, data files, code docs, and even relevant third-party links. These contain detailed insights that may not all be published on the website.
By auditing and gathering these sources, you get a comprehensive pool of company knowledge.
Defining Roles: Website Knowledge Base vs. AI Vector Store
Even after centralizing the content, it’s wise to present it in two complementary forms, each with a clear role:
The Website Knowledge Base: This is a curated set of articles on your site that provides high-level guidance and answers to common questions. Think of it as the user-friendly handbook – it contains overall explanations, step-by-step instructions, and FAQs that customers can easily read and understand. The KB shouldn’t overwhelm users with every detail; instead it offers the essential information and steps to guide users through common tasks or issues.
The AI Vector Store: This is an AI-oriented knowledge repository holding detailed and unstructured information in vector-embeddings for semantic search. It acts as the AI’s brain. All those support notes, detailed technical docs, data tables, and code snippets are ingested into this store (after processing) so that the AI can retrieve fine-grained information when answering user queries. In essence, the vector store is a robust, centralized database of knowledge drawn from all your sources (product docs, tickets, user feedback, etc.) that serves as the foundation for AI analysis and Q&A.
By separating these roles, you minimize overlap between what lives on the public KB and what’s hidden in the AI’s index:
The KB articles cover the most common scenarios in a concise way (no need to duplicate lengthy technical text there).
The AI store holds the in-depth information. When users ask the AI for more detail or edge cases not on the site, it can draw from this rich repository to give an answer.
This way, users have two paths: a self-serve KB for quick answers and an AI assistant that can provide deeper insight on demand. They work in tandem: the knowledge base builds confidence with well-structured guidance, and the AI can fill in details or handle unique questions.
Additional Considerations and Best Practices
Maintain and Update the Content Regularly
Feeding your AI outdated or incorrect information can be worse than providing no information. Make sure that whenever a process changes or a new requirement emerges, you update both the public article and its hidden notes.
Treat the knowledge base as a living document. If you fail to do so, users may get frustrated by contradictory or old answers. Schedule periodic reviews of both visible articles and notes to keep them current. An AI is only as good as its knowledge base – if the knowledge base evolves with your business or organization, the AI will continue to give accurate answers.
Structure and Format for AI Consumption
- The way you format your content can impact how well the AI understands and retrieves it. Use clear headings, subheadings, bullet points, and other structured formatting in both articles and notes. This not only helps human readers scan the article, but also aids the AI.
- Semantic structure (like proper HTML headings or markdown) can improve the AI’s ability to find the relevant section of an article when a question is asked. For instance, having a section titled “Requirements” in your hidden notes, or an FAQ list in the support notes, means the AI can pinpoint that section when the query matches those contexts.
- Avoid burying critical info in long paragraphs without structure – breaking them into lists or Q&A format can make retrieval more precise.