July 23, 2019
TABLE OF CONTENTS
Artificial Intelligence (AI) is the future—at least that’s what solution providers are persistently saying. The real challenge lies in finding out how software companies are actually incorporating the technology into their solutions. Even if AI and machine learning (ML) are actually being used in a substantial way, does it always have a real impact on the business process?
One common example of leveraging AI and ML in the Accounts Payable (AP) process is in the data capture process. Optical Character Recognition (OCR) has been used in the back office for years; in AP, it’s used to ingest invoices in formats such as paper, email, and PDF and transfer the data into searchable, modern, electronic Invoice format. However, it’s rarely accurate on 100% of the documents it scans, and often requires human intervention to correct errors and complete missing fields.
By leveraging ML within OCR technology, solution providers intend to rectify the shortcomings of OCR to make it more adaptable and accurate, as ML will “learn” the format of invoices over time and effectively improve the accuracy rate of the OCR technology. The value of this tool often depends on the level to which the solution provider has integrated ML into the OCR technology, and how much work and time it requires from the end user to significantly reduce exception management for AP teams. This ML/OCR learning curve affects ROI timeframe of a data capture tool.
Paramount WorkPlace is a software provider that is trying to address this learning curve. With their latest product update, Paramount WorkPlace incorporated AI and machine learning into their automated invoice capture tool. In preparation for this release, they exposed the tool to thousands of example invoices so as to train the machine learning to recognize different types of invoice formats and information.
This potentially means that when the client installs the tool, there is close to no setup required for the OCR to successfully capture the basic invoice components right away, such as line items, vendor name, item quantity, or price. Paramount showed us an example of their data capture tool in real time during a recent product demo, uploading an example invoice with errors, logos, and ink smudges, and successfully capturing the data that would matter to finance teams.
Even with high out-of-the-box accuracy, ML-based OCR will still need to improve, and it is important that the AP software allows human users to “teach” the technology. The WorkPlace tool builds up a profile for each vendor’s invoices as it learns, and users have the ability to constantly refine these profiles. Users can correct the WorkPlace tool if the automated invoice capture misinterprets items on an invoice, and adjust vendor profiles based on unique factors.
For example, if certain vendors use odd terminology to refer to quantity, price, or other elements, the user can show the OCR what the vendor is referring to. The system then incorporates that information into the vendor’s profile, and the next time an invoice from the vendor is uploaded, the tool is better able to capture the data.
One great complaint against OCR technology has been that it doesn’t really get rid of the invoice data entry problem that plagues many AP departments–it just makes it a digital process rather than a paper one. By leveraging ML in the OCR tool, an AP software actually can free up AP teams to do more strategic activity. These types of solutions also potentially minimize implementation and training time that come with many more dated invoice capture products, and leads to a shorter ROI timeframe for buying companies.
At the moment, Levvel Research sees this level of advanced OCR mostly in solutions catered to the enterprise level segment, and it’s not often found in tools focused on the middle market. The middle market is also considerably more cost-sensitive than their enterprise counterparts, so the Paramount WorkPlace (a company that typically targets mid-market revenue segment) data capture tool could be could be particularly appealing to companies in this market segment.
If you’re interested in learning more on how software providers are incorporating advanced technology in AP automation, please read our 2019 Levvel Research Payables Report.
Major Bottoms Jr.
Major Bottoms Jr. is a Research Consultant for Levvel Research based in Charlotte, NC. He plays a key role in the analysis and presentation of data for Levvel’s research reports, webinars, and consulting engagements. Major’s expertise lies in the Procure-to-Pay, Source-to-Settle, and travel and expense management processes and software, as well as technologies and strategies across DevOps, digital payments, design systems, and application development. Prior to joining Levvel, Major held various roles in the mortgage finance field at Bank of America and Wells Fargo. Major graduated with a degree in Finance from the Robert H. Smith School of Business at the University of Maryland.
2020 has been an eventful year for Travel & Expense Management (TEM). Because of the COVID-19 pandemic, travel has declined significantly, which made it seem like TEM automation would take a back seat to more urgent priorities.
Although COVID-19 will eventually reach a resolution, the agility that investment in technology empowers will endure. And a key tech investment organizations can make is adding software to their back-office processes.
Experts from Levvel and Centene discuss insights gained from our 2020 Insurance Technology Executive Report. We cover top challenges of insurers today, industry themes, ways to benchmark your organization, and how to get ahead.
This guide explores the potential ROI of automating core Procure-to-Pay processes and enabling a remote workforce for the new business environment.