The swift rise of chatbots in the enterprise has not gone unnoticed by Hewlett Packard Enterprise Co., which today is rolling out a new release of IDOL, its data navigation engine, that adds a natural-language query mechanism for easing searches.

The feature enables workers to explore their organizations’ information troves using the same type of plain English commands that the average Slack bot is designed to take. Under the hood, however, requests are processed much differently because of the complexity of enterprise records.

The first stop for every natural-language question in IDOL will be a parsing engine called Answer Server that can decipher its meaning and check the instructions against the available data sources. From there, the command is passed on to three record extractors tasked with pulling requested information from the appropriate systems.

The first and most straightforward of the bunch is Answer Bank, which companies can configure to provide prepared answers for common questions. A retailer, for instance, could have the mechanism automatically surface its latest inventory numbers when a business analyst asks about merchandise availability. It’s also possible to display entire dialogues such as step-by-step instructions for configuring a piece of equipment.

More fine-grained queries are handled by a tool dubbed Fact Bank that can pull specific snippets of information from a company’s records. HPE envisions the feature being for niche requests that organizations can’t address in advance with Answer Bank. A similar purpose is served by Passage Extract, the third data extractor included under the hood, which  is designed to provide overviews about specific topics, events and people within a company.

Organizations can use the new natural-language capability on its own or in conjunction with dialogue generation tools such as the open-source Nadia framework. The latter tool makes it possible to have IDOL followup data requests with various questions about the desired information, such as what specific details should be extracted and from where — in other words, exactly how a chatbot would be expected to behave on Slack or HipChat.

Image via Pixabay