Browse the docs
Concepts
The knowledge base
How ASDK turns your website and documents into one knowledge base that every agent answers from, how answers cite it, and how each site's data is kept apart.
Last updated
On this page
The knowledge base is ASDK's copy of what your website says, organized so that agents can find the right passage for a question in a fraction of a second. You do not write answers by hand: you keep your website and documents up to date, and the knowledge base is built from them.
Every agent on your site uses the same knowledge base. The Chat agent answering a visitor, the Sales agent qualifying a lead and the Agent-Ready agent describing your business to an AI assistant all draw on the same facts.
What goes in
- Your website's pages. ASDK's crawler reads the public pages of the site you connect, the way a search engine does. It identifies itself and follows your robots.txt rules; the crawler page explains how.
- Documents you upload. Product sheets, price lists, policies and other material that is not on the site, or not in a form a crawler can read well.
Each plan includes a number of pages; the limits are on the pricing page.
How it is built
The steps are the same ones most question-answering systems use. The glossary explains each term in more depth.
Read
The crawler fetches your pages and keeps the text that carries meaning. Uploaded documents are read the same way.
Split
Long pages are split into short passages that each cover one idea — a step called chunkingChunking is splitting long documents into smaller passages before indexing, so an AI system can retrieve just the part of a page that answers a question.. A passage keeps a link to the page it came from, so answers can point back to it.
Index
Each passage is turned into embeddingsEmbeddings are lists of numbers that represent the meaning of a piece of text, so software can find related content by measuring how close two lists are., numbers that capture its meaning, and stored in a vector databaseA vector database stores embeddings and quickly finds the ones closest to a query, which lets AI systems search content by meaning instead of exact words.. That lets ASDK find passages by what they mean, not only by the words they share with a question.
How answers use it
When a visitor asks a question, ASDK looks up the passages that best match it and gives only those to the language model, with instructions to answer from them. This pattern is called retrieval-augmented generation, or RAGRetrieval-augmented generation (RAG) is a technique where an AI system first looks up relevant documents, then writes its answer using what it found..
Two rules follow from it:
- Answers cite your pages. The agent shows which pages an answer came from, so visitors can check it and keep reading.
- No passage, no answer. If the knowledge base has nothing relevant, the agent says it does not know and offers a way to contact you, instead of filling the gap with a guess. The glossary entry on groundingGrounding means tying an AI model's answer to specific, verifiable sources, such as your own pages or a search index, so each claim can be checked. explains why this matters.
Because every answer starts from your content, the best way to improve answers is to improve the content: a clear pricing page, specific product pages and an honest FAQ help visitors, search engines and AI agents alike.
Keeping each site's data apart
Each site's knowledge is stored separately from every other customer's, and agents on your site can only search your site's knowledge. Your content and your visitors' conversations are used to serve your site and are not used to train foundation models. The security page describes the safeguards, and the privacy policy covers personal data.