IBM is rolling out a cognitive tool powered by its Watson supercomputer that uses companies' internal data to answer questions about performance and efficiency, and even make predictions. Watson Analytics is a software service, delivered over the cloud. It is a result of a collaboration between teams from IBM’s data analysis group and its Watson unit, which has been built into a business in the three years since Watson beat human champions in the question-and-answer game "Jeopardy."

The ambitious project aims at taking to take a big step toward getting analytics in the hands of every business user," according to IBM.

IBM has shown an early working version of Watson Analytics to a handful of customers and industry analysts. It combines basics of data handling with the Watson technologies of natural-language processing and machine learning. A result, they say, is that a business person, who is not a statistician or data scientist, can type in questions to probe corporate data. Watson Analytics clients can upload their business' data and ask the tool personalized questions like, "Who are my most profitable customers?"

Watson, which uses artificial intelligence to quickly analyze huge amounts of data and can understand human language, may ask several follow up questions to give clients the most accurate results possible. The cloud-based tool will then analyze various data sets and provide answers and visualizations that predict future outcomes. It can also create graphics and charts that allow clients to share the outcomes of their inquiries and better implement solutions to their problems.

Watson Analytics may respond with answers ranked by probabilities or with suggestions. The suggestions can begin a sort of dialogue, recommendations to improve the data used or add other sources. It can, for example, rate the quality of the data probed by an initial inquiry. More reliable predictions, the Watson software might say, would result from cleaning the data or including other sources, like local weather or traffic flows, for retail sales predictions, for example.

The offering will be available for all IBM business clients in November.