AOL’s Area 51 incubator releases first prototype — a bot for its ad tech data
Called Avery.ai, it now moves into product development and joins other companies’ efforts to make marketing data more readily accessible.
For most of this planet, Area 51 is known as a place where engineers may be unraveling some out-of-this-world technological secrets.
For AOL, Area 51 is the name of an incubator where selected employees and recent college graduates create new tech, with six months of funding and access to AOL resources. The incubator has now announced its first prototype: a Slack-integrated chatbot that allows marketers to employ natural language for querying data in the AOL ad tech platform.
Called Avery.ai and developed by AOL entrepreneur and researcher Davood Shamsi, the prototype is now being handed off to AOL’s business units, in the hopes that it can be turned into a product.
A user can type an unstructured request like “pull hourly impressions in July 2016.” It currently takes about 20 to 30 seconds for the bot-in-development to respond. And, at least in its current form, it quaintly addresses you as “boss.”
Shamsi told me that, while Avery is tuned for AOL’s ad tech environment, it can utilize any data source, structured or unstructured. He added that it might be ported to platforms other than Slack.
CTO Bill Pence pointed out that many analytics front-ends suffer from what he called “dashboard fatigue,” where users have to figure out how to find what they need and are limited by the dashboard’s choices.
By contrast, he said, Avery figures out how to fulfill your request, and it can handle an unlimited variety of requests.
Typically, Shamsi said, the bot’s requests to the AOL ad tech platform are about events-based data, such as campaigns. A user might want to compare two different campaigns, for instance, which a dashboard might not allow.
He added that, in testing, Avery answered about 40 percent of queries after they are asked once, and another 40 percent when the original queries are rephrased. Of the remaining 20 percent, he said, about 15 percent will be answerable with further development, and perhaps 5 percent will be beyond the bot’s current capability.
Avery.ai picks up on a recent trend to make data queries much simpler than they have been, by going beyond the ubiquitous dashboard.
Last year, for instance, New York City-based business intelligence provider Sisense released three kinds of non-dashboard interfaces for its data: an Amazon Alexa skill (that is, an app), a chatbot and a connected light bulb that shines a different color if a given threshold of data is breached.
Similarly, marketing analytics platform Datorama launched an Alexa app in December. But AOL’s Pence noted that he hasn’t seen anyone create a bot like this specifically for ad tech data.