The name describes the target format, not a different technology: an answer engine gives back one direct answer, sometimes spoken, sometimes a single AI Overview box, instead of ten blue links to choose from. Content written for that has to work as a complete answer on its own, because there is no surrounding page for the reader to scan for context.
How it relates to GEO
AEO is usually treated as the conversational-search-specific end of Generative Engine Optimization: GEO covers the broader goal of being found, verified, and cited by any generative AI system, while AEO focuses specifically on the direct-question-and-answer format, FAQ pages, structured Q&A, self-contained definitions, that a voice assistant or chat interface can lift and quote whole.
What it looks like in practice
Front-loading a real, quotable answer in the first sentence, using FAQPage structured data so the question-and-answer pairing is machine-readable, and writing definitions that are true and complete without needing the rest of the page for context. Kelriva's own glossary is built on exactly this pattern.