Unlocking business potential by transforming Data Extraction Processes with Hyperscience.
Since late 2017, Hyperscience has been working alongside a large financial services company in the United States that offers financial products and services to individuals and institutions through financial intermediary firms.
The company was looking to achieve digitization of service across two fronts. One was championing true digital solutions that tied into online portals or helped deliver and totally digitize an otherwise traditional paper-based activity, and the second was trying to automate existing manual processes.
We asked ourselves how do we try to put automation in place that helps reduce manual activity and speed up that execution with high rates of accuracy? The three components were mitigating costs, delivering with high quality, and doing it as quickly as reasonably possible.
The company started with a complex transaction that requires a lot of customer information and reviews. Hyperscience deployed in a test environment, running through several forms, while the firm engaged an industry consultant to help assess the results and compare to other vendors.
In the spirit and context of adopting an agile delivery approach, a primary goal was to reduce the time it took to deliver, update and maintain the system, so the company needed a tool and a vendor that could provide frequent updates without major headaches or changes and a solution that could be updated with minimal effort.
A majority of what the forms processed was still handwritten, so robust, reliable handwriting and machine print recognition was essential. Hyperscience’s Machine Learning-driven approach – “the idea and objective that as the system processes more of the same forms, there would be improved performance and reduced manual validation” – was fundamental.
Hyperscience Performance Exceeded Expectations
During the testing phase, Hyperscience was able to systematically collect roughly 90% of the fields at 98% accuracy rates out-of-the-box. From a machine transcription perspective, this was much better than what had been previously achieved.
In addition, while the firm had optimized their data keying operations for complex transactions, they weren’t extracting all data on the page. “When we looked at volume potential based on an objective to automate the processing…with Hyperscience we were tripling the number of fields we needed to capture, and to do that at 90-95% machine transcription with 98+ percent accuracy rates was extremely attractive.”
Results & Relationship Today
The company started with high-volume forms where they anticipated the biggest benefit and saw the greatest need and today is collecting twice as much data as a year ago. In addition, they’ve unlocked automation in those business groups that has resulted significant savings in manual effort and time available to focus on more complex work.
Now several years into the partnership, Hyperscience has garnered significant attention across the company, not just for its differentiated features and offering, but for how it’s been deployed by the team in a truly agile approach. “It’s a good representation of what the company and the company leadership was wanting to have happen three years ago. We want to deliver iteratively, we want to take risks, we want to try things and then figure out how to scale. It forces us to think and actually behave differently.”