Six AI/ML services you can use with the Box Skills Kit

With the Box Skills Kit, you can easily connect third-party AI/ML services with Box to process and analyze your business’ content. Our framework makes it easy to fetch files from Box, analyze them using a third-party AI/ML service, and write the output back to Box as metadata. With the rich information extracted from these files, you can make it easier for people in your business to find and work with their content, automate manual data entry, or even automatically trigger security policies based on the contents of a file.

As we gear up to make the Box Skills Kit generally available later this year, we’ve been looking at some of the different AI/ML services that you can use with the Box Skills Kit. Here are a few to consider:

IBM Watson NLU for document insights

One of the more exciting promises of AI/ML is the ability to understand unstructured data at an entirely different level. With natural language understanding, you can understand an extracted text in ways never possible before, identifying concepts, entities, emotions, keywords, and much more. Watson Natural Language Understanding offers powerful algorithms for identifying and labeling these features, even correlating multiple documents to identify similarities. With the Box Skills Kit, you can use Watson Natural Language Understanding to extract information and better understand the contents of text-based files.

eBrevia for contracts analytics

Many legal departments use Box to store and manage contracts securely. eBrevia offers powerful pre-trained natural language processing algorithms that enable you to analyze contracts, such as customer contracts, vendor agreements, employment agreements, patent agreements, and more, that your company may store in Box. eBrevia’s algorithms can extract critical information such as terms, renewal dates, pricing, and much more. With the Box Skills Kit, you could create a Custom Skill that analyzes your company’s contracts to extract pertinent information to help your legal team efficiently manage contracts.

Google Cloud AutoML for custom image recognition

One of the most obvious use cases for the Box Skills Kit is to label objects in image files that your company stores in Box, such as stock photos your marketing team might use on your company website. Images are among the most popular types of content in the enterprise and become more and more popular as the way we work becomes more visual. Cloud AutoML lets you build and train custom machine learning models, including custom computer vision models. Using Cloud AutoML, you could create a custom computer vision model that recognizes specific objects and entities unique to your business, like specific products your company sells or specific logos and sub-brands. Then, you could connect this model to Box using the Box Skills Kit and automatically label your images in Box with tags for those objects and entities.

Leverton for lease abstraction

Let’s say your company is a retail brand looking to open several new brick-and-mortar locations in the next few months. One of the ways you might use Box is to have a secure environment to share and collaborate on lease agreements with local property managers and real estate developers. Leverton offers pre-trained deep learning-based algorithms to automatically extract essential information from lease documents, such as landlord name, rent amounts, or termination rights. By integrating Leverton with Box, any leases uploaded to your company’s instance of Box would be automatically structured, with all the information you need automatically extracted from those documents and stored as metadata. The Leverton platform also offers several other pre-trained algorithms for other types of documents, like mortgage documents and title insurance documents.

VoiceBase for speech analytics

Businesses typically invest in expensive and manual transcription services to better understand their audio. However, converting speech to text is only the beginning of what’s possible for analyzing audio with AI/ML. VoiceBase, for example, offers several capabilities on top of speech-to-text transcription for analyzing audio files even further. VoiceBase’s algorithm can identify any keywords or topics that appear in a transcript, identify and label PII, PCI, SSN, and swear words, and even add “predictions” to determine the outcome of an audio recording, like detecting if a caller made an appointment. You can also create a custom vocabulary to help VoiceBase identify words or phrases that are unique to your business. With the Box Skills Kit, you could connect VoiceBase to process audio content that your business stores in Box to extract rich insights.

Ephesoft for document classification

One of the biggest challenges facing IT organizations is classifying different types of documents. Teams in lending, banking, insurance, healthcare, government, and several other industries spend inordinate amounts of time manually labeling documents, wasting valuable time and resources. Technologies like Ephesoft offer classification algorithms, based on deep learning models and massive training datasets, to classify common types of documents to eliminate manual tagging and routing. Using Ephesoft, you could process documents as they are uploaded to Box and automatically classify them based on the type of document that Ephesoft recognizes.

Acuant for identification verification

For any onboarding process, whether opening a new bank account or getting a new employee set up for your company, you’ll likely need to collect a government-issued identification document, like a drivers license or passport. Because of the sensitive nature of these documents, often stored in image format, many organizations use Box to store them securely. The challenge, however, is that organizations then need to verify these ID images and pull information off of the images. Acuant offers machine learning-based algorithms to analyze images of government-issued identification documents that you could use to structure this content. Your Custom Skill could retrieve images of IDs whenever an employee or customer uploads one to Box, verify the document and extract the individual’s information off of the image to store as metadata. From there, that metadata can organize these images based on specific fields, like geographical location or gender, or go to another system, like a CRM or HCM system, via the Box API.

Get started with the Box Skills Kit

Got an idea for a Box Skill? You can check out our beta for the Box Skills Kit to use these different AI/ML services — or any other service — to create Box Skills. To learn more, please visit our developer website.