In September, Wolters Kluwer was granted a patent for the AI technology that drives LegalVIEW® BillAnalyzer. In this post we talk to Abhishek Mittal, VP of Data Analytics and Operational Excellence for Wolters Kluwer’s GRC division, about the patent and its significance.
What does it say about LegalVIEW BillAnalyzer that its AI technology is patented?
Traditionally, in the legal bill review space, companies have used third-party or in-house legal experts. What we uniquely offer is a combination of advanced technology and domain experts to deliver significantly better results for our customers. The patent supports this approach and indicates that there is something exceptional in what we offer. We have always been proud of our domain experts, but this patent is an external benchmark showing that our technology is also leading the industry and is proven to be unique and cutting-edge.
Not every technology product is patented. How did you decide that we should seek a patent for the BillAnalyzer AI?
I have been involved with developing advanced tech for 15 to 20 years, including in previous roles before I came to Wolters Kluwer. I know from experience that AI technology has been prevalent in aspects of legal operations for some time, but what is unique to LegalVIEW BillAnalyzer is the specific way we apply AI technology to bill review. We are able to do this because we have access to data that gives our AI models an advantage.
In addition to having the LegalVIEW database to power our AI, we are also the market leaders with data and experience based on countless transactions across hundreds of customers. Our AI can learn from this unique combination of advantages, and we were the first company to introduce a solution of this kind to the market. These combined strengths gave us confidence that we could make a successful case for a patent.
What is it about the AI technology used in LegalVIEW BillAnalyzer that sets it apart from other AI technology products?
Our data is key, but there is some nuance to that idea. People often think it’s as simple as “more data is better.” That’s true to a degree, and in LegalVIEW, we do have the world’s largest database with over $140 billion in legal transactions. But the quality of the data is also very important. Ours is carefully curated data. It has been vetted by legal billing experts to identify, for example, which line items require adjustment and which don’t, which charges are billing guideline violations and which are not. That level of quality and curation is not easy to achieve. You could have the best data scientists in the world, but if you don’t have access to that kind of labeled data set (and no one else does), your model will not be as good as ours.
That’s the strength of the data and of the AI itself. The combination of AI with the workflow of our market-leading ELM platforms and e-billing experts allows us to deliver an end-to-end solution. That eases the change management for our customers and is our engine for differentiation.
What makes AI so well-suited to the bill review process?
There are a few traits that make it a good fit. First, bill review offers a large number of transactions with similar behavior. You have millions of line items and billions of dollars, all going through a very similar process. This sets the stage for success with AI because there is more than enough data for the models to learn from. Second, there is too much data for humans to be able to reliably recognize patterns in it. But for AI, more data makes pattern recognition better. Machine learning can identify patterns across law firms, work areas, and timekeepers that human reviewers miss. And third, bill review provides us with a feedback loop in which a billing expert looks at the line items flagged by the AI and indicates whether each one does or does not need adjustment. This feedback allows the AI to improve constantly.
When developing this AI technology, did you work with clients to get their feedback as you progressed?
Yes. This is not a “one and done” type of technology. We have had multiple iterations of the AI models with continuous enhancements based on input from customers. This is the only operational process I’ve seen where the operations team has met with clients every week for years. We have learned so much across different industries and segments; and we’ve been able to use all of that to improve the models tremendously over time.
Why is bill review better with AI than without it?
The AI acts like a highlighter to prioritize the line items most likely to need adjustment. It helps the bill reviewers focus on the biggest opportunities for increased compliance and cost savings. This has a huge impact on maintaining good results even as reviewers get tired. We did an experiment in which we tested expert reviewers with and without the help of AI. Early on, the results were similar. But as the reviewers got fatigued, those without AI assistance would often focus their remaining energy on the highest value line items. This is understandable, but we found that they missed a huge amount of adjustments in the lower value line items. Meanwhile, the AI users captured those adjustments and got better results. The AI improves focus and helps prioritize in a way that directly adds value. This has always been an advantage with our BillAnalyzer Expert Service. Now that we’ve introduced the BillAnalyzer Data Service, clients can get this important benefit for their in-house reviewers, as well.