News Stay informed about the latest enterprise technology news and product updates.

Big data meets automated content development

I was thinking about automated content development before it was possible. Now it’s a reality.

My first job in journalism was to take calls on Friday nights from high school football coaches, who would relay the stats and a few details about the game. A bunch of other college interns and I would take the information, create a box score and write a short story about the game, which would then get compiled into a conference “wrap.”

The job was interesting but so repetitive that I used to say that robots could do it. It turns out I was right.

It’s being done now by Automated Insights, a Durham, N.C., startup. The company has married big data and content creation with developers and writers to create completely automated news and information services. I met with founder & CEO Robbie Allen at the MIT Sloan Sports Analytics Conference, and he explained how he started out in sports but is now looking to branch out.

“We have done a lot in sports, but we’ve done a lot in other verticals as well, like finance, real estate and traffic, and others where there are large sources of data,” Allen said. “We are moving forward with the approach that we are not just sports, but a B2B company that’s leveraging customers’ big data to generate unique and interesting content that they can use in a variety of ways.”

I should be worried that computers have finally replaced real writers, but when you consider the demand for content around large data sets (think fantasy sports or rush-hour traffic reports), there is a real need.

Automated Insights’ writers design what the content is going to look like and the developers translate that into code, Allen explained.

“The writers determine what are the things that go into a game recap; what are the data triggers that we would want to generate a sentence or paragraph about,” he said. “And the developers are responsible for marrying that design with the available data to generate actual content.”