In recent years, many research systems have been proposed to perform data extraction and automation tasks on Web sources. Since most of today’s Web sources are “human-readable” but not “machine-readable”, these systems must address a number of difficult challenges, such as dealing with complex navigation sequences, extracting data from HTML pages and reacting to source changes. Denodo Corporation has developed ITPilot, an industrial-strength solution that allows complex “wrappers” for Web sources to be graphically generated and automatically maintained. This paper presents the architecture and the basic ideas “behind the scenes” in ITPilot.