Traditional databases give users little help in terms of accessing the database itself. They are searchable by keywords and phrases that are connected by Boolean operations such as AND, OR and NOT. Intelligent database interfaces, however, are cache-based, and are designed to efficiently access one or more database management systems (DBMSs), remote or not. As a result, they provide a lot of choices and flexible options for conducting queries. For example, when a user types in a phrase in a library search computer, the intelligent database then provides a lists of results that are arranged according to how related they are to the phrase. The AI may also correct some misspelled words or grammatical errors entered by the user. Some databases also display synonyms or antonyms for the keywords and phrases. This search technique has been used at Google since its infancy.

This concept assumes three levels of intelligence in the system: high-level tools, a user interface and a database engine. The high-level tools layer, which often uses artificial intelligence techniques, manages the quality of data and automatically discovers relevant data patterns with the help of a process called data mining. The user interface level uses hypermedia, which uniformly arranges text, images and numeric data. Finally, the intelligent database engine serves as the foundation for the other two layers, often combining relational database (RDBMS) techniques with object-oriented ones (OODBMS).