Barely a week seems to pass without a web company open-sourcing some new tool for building artificial intelligence software. Last Thursday, it was Facebook Inc.’s turn. The social networking giant released a framework called Torchnet that contains five ready-made modules for training and testing deep learning algorithms.

The modules are implemented in the form of interoperable programming abstractions that can be added to an artificial intelligence project with relatively little modification. Facebook says that adapting an algorithm to ingest a new dataset, for instance, requires changing just a single line of code in Torchnet. As a result, developers are able to move onto other tasks faster and meet their delivery deadlines more easily.

Reducing the amount of manual work involved in enterprise technology projects is also a focus for Microsoft Corp., although it’s approaching the issue from a different angle. The software giant unveiled a new cloud service last week that promises to automate the enforcement of corporate data governance policies by harnessing textual analytics. Azure Information Protection, as the offering is called, can scan Office 365 documents for sensitive details like credit card numbers and adjust access permissions accordingly. Restrictions applied by its algorithms are enforced even if a file ends up outside the corporate network in a file locker or on a worker’s mobile device.

Oracle Corp., one of Microsoft’s biggest rivals in the analytics market, also managed to make headlines last week after launching a pair of new database appliances designed to make its relational store more accessible to SMBs. The X6-2S is a single-socket system that can be ordered with up to 384 gigabytes of memory, while the X6-2M provides up to 768 gigabytes. On the storage front, meanwhile, both offer a maximum capacity of 12.8 terabytes.

Image via Pixabay