Resources
Where to start, depending on what you are trying to work out.
Start here
If you are evaluating this category rather than us specifically, three pieces cover most of what matters: the vendor evaluation checklist, the piece on what visitor questions reveal, and the explanation of why retrieval quality dominates model quality.
For website owners
The content gap cluster covers reading questions as a content map, turning transcripts into a roadmap, and the specific questions pricing pages fail to answer. The deployment cluster covers what has to be true before an assistant is worth having, and platform specific notes for Webflow and WordPress.
For engineers
The retrieval cluster covers chunking, embeddings and what similarity search cannot do. The reliability cluster covers provider failover, rate limits, refusal design and latency budgets.
On data and trust
The data cluster covers where content goes when you index it, how to decide retention rather than copy it, and what you can honestly tell your visitors.
Definitions
The glossary defines the terms that appear across these pieces, including retrieval augmented generation, embeddings, chunking, grounding, hallucination and deflection rate.