Data integration provider SnapLogic Inc. is moving into the artificial intelligence market with a new toolkit for building and deploying neural networks.

Launched today, SnapLogic Data Science expands the company’s flagship Enterprise Integration Cloud, which enables enterprises to link internal applications with one another using customizable connectors called Snaps. These Snaps make it possible to share information among otherwise disjointed systems.

With SnapLogic Data Science, companies can now use machine learning to analyze the information that flows through Enterprise Integration Cloud. The toolkit includes Snaps with built-in AI models capable of making predictions based on historical data and performing classification tasks such as image recognition.

Organizations with more advanced requirements can develop their own models. A second set of Snaps provides the ability to generate a neural network from one of several ready-made models, customize key settings and then train the AI using internal company data. The toolkit also includes features for performing cross-validation, an extension of the training phase focused on fine-tuning AI accuracy.

The core machine learning Snaps are joined by several other connectors designed to ease the task of readying information for analysis. SnapLogic said that developers can use them to automate common preparatory steps such as converting records into a new format, as well as extract statistics and metadata from records.

All these tasks can be carried with “minimal” coding, according to the company. Since SnapLogic Data Science is part of Enterprise Integration Cloud, it lets developers use the platform’s visual interface to build their models.

In an interview on SiliconANGLE’s theCUBE studio earlier this year (below), SnapLogic Chief Scientist Greg Benson explained that the company provides a graphical interface that lets users drag and drop Snaps. “You can put them together like Lego pieces to define sophisticated tasks so you don’t have to write Java code,” he said.

Alongside the drag-and-drop capabilities, SnapLogic Data Science offers a connector called Remote Python Script Snap that targets more code-heavy use cases.  Python is one of the most widely used programming languages in the machine learning ecosystem. The connector enables companies to use external AI development frameworks such as TensorFlow and Python machine learning libraries.

The toolkit isn’t SnapLogic’s first foray into AI. Integration Cloud features an existing recommendation engine that uses historical data to generate advice on how to carry out development tasks.

“We use machine learning to help you build out these pipelines so we can anticipate, based on your data sources, what you are going to need next,” Benson said. “That lends itself to rapid building of these pipelines.”

Photo: SnapLogic