Richly funded data integration startup SnapLogic Inc. says it’s building artificial intelligence into its cloud-based service, potentially cutting months of time off of large-scale development projects.
Named Iris, after the mythical character who was the messenger of the Greek gods (pictured), the proprietary machine learning technology is the first in what SnapLogic says will be a series of increasingly intelligent improvements to its service that take advantage of millions of metadata instances and data flows it has captured in its Enterprise Integration Cloud.
Its initial form, Iris will take the form of an integration assistant that recommends next steps in building data pipelines. Chief Executive Gaurav Dhillon was careful to compare the assistant to the search engine feature that automatically completes partial queries rather than to Clippy, the much-hated virtual assistant that plagued Microsoft Office users for several years.
“If you’re trying to connect SAP to Workday, you drag in a snap and it tells you what to do next,” said the CEO. Snaps are the connectors SnapLogic uses to create data pipelines. The company said the technology will be increasingly integrated into its core service over the next three years to automate large parts of processes that are currently manual. Future releases will be better attuned to application usage with the eventual goal being to “put down endpoints and say ‘show me the way to do this,’” Dhillon said. “Our vision is to take this to the autonomous driving level.”
Hosting its service in the cloud gives SnapLogic an inherent advantage over other integration vendors, many of whom are rooted in packaged, on-premises software, Dhillon asserted. SnapLogic can use workflows captured by its service to train machine learning models that continually improve over time. Customers are also tending to use fewer integration endpoints as cloud markets consolidate, making the training process simpler. “There’s a very high degree of repeatability because there are now just a few big SaaS companies and three horsemen in the cloud,” he said, referring to software-as-a-service and the major cloud providers, Amazon Web Services, Microsoft Azure and Google Cloud Platform.
SnapLogic claims that it is seeing 80 percent to 90 percent accuracy rates so far on recommendations and that it expects accuracy to approach 97 percent by the time the capability is released on May 17. The technology was developed by an internal team led by Greg Benson, a computer science Ph.D. who’s SnapLogic’s chief scientist and a professor at the University of San Francisco. It represents several “person-years” of development, according to Dhillon.
The features will be included as part of the basic SnapLogic service at no additional cost. The company has raised more than $136 million, including a $40 million round in December, and has made no secret of its aspirations to be publicly traded.
Dhillon spoke to SiliconANGLE Media’s video studio theCUBE at the Big Data SV event in San Jose in March: