At Hyperscience, we’re constantly thinking about how our solution can better serve our customers’ needs. This is what drives our organization, and whether we’re on the sales, operations, customer success, engineering or product teams, we work hard to gather feedback from users in the field and continuously improve our platform’s functionality and user experience. These improvements are packaged and released quarterly. In July, we shipped version 25.0.0. Below, we’ve rounded up some of our favorite enhancements.
With our latest release, we launched a refreshed UI with a left-side navigation bar, a new workflow to organize documents together, and French language support.
When we make product and design decisions, we solve for the root of our customers’ pains, beyond what we know already exists in the market. Our guiding principle is to distill complexity and cater to non-technical teams and business stakeholders. Our intuitive left-side navigation, for example, allows users to quickly jump from completing lightweight manual exception handling (simple tasks that train our Machine Learning models) to analyzing our reporting dashboards that display Hyperscience’s unique ability to measure its own accuracy and automation. Our new document organization workflow accommodates how our various users receive documents, categorize them, and push extracted data downstream, processes that vary greatly by customer.
With our support for French, we took the first step towards adding three additional languages by the end of 2019. Our ultimate vision is to make our platform input-agnostic, capable of extracting data from every imaginable document (i.e. any structure and language), and flexible enough to adapt to any processing workflow.
More Machine Learning support
Hyperscience’s 25.0.0 release also includes specialized models to maximize extraction automation and accuracy for checks (in addition to invoices). This is important for our customers across industries whose downstream processes require quick and accurate processing of checks, whether it’s deposits to a brokerage account or straight-through processing of high volume, low complexity medical claims.
We also improved data standardization for currencies, legal amounts and numbers with a unit. For example, customers might write the date as 12-7-19, 12/07/19 and Dec-7-19 all on the same form. Hyperscience not only extracts that data, but structures it in a standard format, so all can be read as MM-DD-YYYY (or whichever date format is preferred). In addition, by reading documents with context the same way a human would, our solution knows that “four hundred and fifty dollars” is the same as $450.00.
Additional input points and connectors
In simplest terms, our solution takes a PDF or image file (e.g. PNG, JPEG, TIFF) and extracts data. How our solution ingests that PDF or image, however, depends on a customer’s existing workflows. Version 25.0.0 introduced a growing number of queues, folders, and RPA providers, including IBM Message Queue, RabbitMQ, and UiPath. In addition, the Email Listener input connector can connect to any email inbox and create submissions from email attachments or the email body. What does this mean? Imagine, for instance, that a customer attaches a video file to an email they’re sending their insurance provider about a property claim. HyperScience wouldn’t process the video, but our enhanced submissions/document output page will indicate that there was an additional file attached to the email, putting important details in one central place in case it becomes relevant to downstream processes.
All of this means that users can spend less time implementing Hyperscience and more time leveraging the benefits of intelligent document processing. Better downstream business decisions start with high quality data extraction, and our connector with UiPath, for example, allows customers to extract text from previously inaccessible PDFs or images and easily drop structured data into existing UiPath workflows.
It used to be that enterprise software required significant developer resources to set up, implement and maintain. Even today, legacy data capture products can take months and hundreds of hours of development to get new use cases up-and-running, which makes bringing on new business lines challenging. But enterprise software and easy-to-use are no longer mutually exclusive, and we’ve proven that.