PwC helps make compliance easier, automates regulatory obligation identification with Microsoft Azure Cognitive Search

Global professional services network PwC is known for innovation and quality results. PwC is a network of business consultancy and accounting firms in 150 countries, making it the largest professional services organization in the world, according to the Reference for Business. 

Regulated companies like PwC’s Fortune 100 clients can invest a lot of time and effort to identify and understand the regulations that they must meet. To help its clients comply more easily, PwC combined its AI expertise with Microsoft technologies. The result is the Regulatory Obligation Identifier (ROI), a proof of concept (POC) of a knowledge mining solution that uses Microsoft Azure Cognitive Search and Azure Functions to add an easy way to surface predictions from PwC’s deep learning model. PwC’s ROI enables Azure Cognitive Search customers to apply AI models that automate regulatory data identification to their document data. Compliance is getting easier, cheaper, and faster.

Seeking a faster, better way with AI

Knowledge mining is an emerging category in AI. It uses a combination of AI services to help drive content understanding over vast amounts of unstructured, semistructured, and structured information. That translates to powerful capabilities for businesses: to explore their information, uncover insights, and find relationships and patterns at scale. 

Azure Cognitive Search is a cloud search as a service with built-in AI capabilities that enrich all types of information, easily identifying and exploring relevant content at scale. Developers use it to compose a set of skills (prebuilt functionality that they can add with a simple checkbox, like language detection, translation, sentiment analysis, key phrase extraction, and entity extraction). 

Microsoft partners can easily extend the capabilities of Azure Cognitive Search with their own AI models, like the PwC ROI solution. 

Finding the needle in the haystack

Large companies face a perfect storm of compliance risk. Regulations are many, dense, and constructed in complex legal language. They issue from a number of regulatory agencies, with probable new rules on the way related to the 2020 pandemic. Compliance for global companies is further complicated by different sets of regulations in each country where they do business, issued in a variety of languages. Penalties are high. And media reports of noncompliance can lead to a public relations nightmare.

A company can accidentally violate a regulation because of the difficulty of unearthing the applicable section. It can take hours or weeks for one experienced professional (read: expensive) to find the information, and the process is error prone. Apart from the expense, companies would much rather focus on how to meet those regulations strategically. 

Neelam Sharma, Director of Digital Risk Solutions at PwC, sympathizes. “Our clients are on the forefront of some of the biggest challenges in the industry,” she says. “They’re trying to comply, and do it faster and cheaper.” When a client who had to distill a regulation under a looming deadline asked Sharma and PwC Machine Learning Scientist–Manager Todd Morrill for help, they looked to AI technology and delivered results in minutes. 

Choosing an AI technology that meets a range of needs

PwC doesn’t dictate platforms to its clients. “We’re technology agnostic,” says Shana Reich, Enterprise Project Management Officer at PwC. “The key drivers for us are how the technology will help support the business need and the organization overall, and how well it will perform. This is why we are helping make it easy to consume PwC AI models alongside the data and analytics services our customers choose, like Azure Cognitive Search.” 

When Morrill considered AI technologies for the PwC ROI solution, he looked for maturity, effectiveness and efficiency, security, and ease of use. Azure was a top pick. “I love to choose which tools I need. On-premises tool sets are more restrictive, and that can create development friction,” he says. “With cloud-based solutions like Azure, developers can have the right tool for every task. The Azure AI platform makes it easy to develop models and Azure Cognitive Search makes it easy for us to bring the models to where customers are working with their data.”

PwC defined regulatory search capabilities as cognitive skills in its ROI solution, making it easy for Azure customers to use them within their Azure Cognitive Search solutions. The PwC ROI solution facilitates custom machine learning skills so that users can simply deposit regulatory documents or PDFs into Azure Blob storage. Azure Cognitive Search performs language detection and translates the content if necessary, then runs it through PwC’s ROI model to find actionable regulations. Other data may also be extracted through this Azure functionality—entities like organization names or locations, for example. Users can store and query all of this extra intelligence from their document at will.  

Ease of use looms large for Sharma. Azure Cognitive Search makes it easy to mix and match the right cognitive skills for the task for easy translation of regulatory documents prior to applying the PwC ROI skill. “Azure is a one-stop place where you can do the work without having to cherry-pick tools from disparate solutions,” she says. “Not only is all that capability and availability included on Azure; the ways that Microsoft has developed Azure cognitive skills means that we can take advantage of work that has already been done.” 

Sharma and Morrill used the ROI solution to help analyze their client’s lengthy regulation—with stunning accuracy. “At first he claimed that our model hadn’t worked because it didn’t map to the parts of the regulation he was analyzing,” recalls Sharma. “Then we all realized that the model had correctly analyzed an entire 100-plus-page regulation in an hour. Our client had spent many hours focusing on just one small area.”

Distilling knowledge and technology into a powerful tool

With Azure Cognitive Search, PwC gained a fully managed cloud search service interoperating with Azure storage solutions and index functionality for lower operational overhead and better search performance. Customers can use ROI to help eliminate manual work and reduce the time for knowledge mining a huge corpus of documents from weeks to seconds. PwC incorporated Azure Blob storage for machine learning workload storage that was scalable. “Blob storage is invaluable. It provides an artifact repository for our datasets and trained machine learning models,” says Morrill. He explains that examining a more than 100–page regulation could result in thousands of individual data points. With Azure Cognitive Search, he can ingest entire data lakes of documents to Azure Blob storage or other services.

Extending efficiency

PwC needs efficient AI technologies that help its developers focus on creative problem solving. Morrill deployed the ROI search model to Functions, eliminating the need to develop an API. “As we’re about to launch our ROI POC, I can see that Azure Functions is a value-add that saves us two to four weeks of work,” he says. “It takes care of handling prediction requests for me. I also use it to extend the model to other PwC teams and clients. That’s how we can productionize our work with relative ease.” It’s cost effective, too. Functions doesn’t start until it’s needed, then ends after the compute completes. Morrill’s team pays only for the compute it uses.

Productivity multiplies over time. “Complying with regulations is continuous, with constant updates. That’s another reason to relegate mundane tasks like this to solutions that PwC and Microsoft can build, like our ROI product,” says Sharma. That productivity boost ripples across the multiple lines of users working with regulatory compliance, from first-line legal researchers to policy and procedure writers to those in charge of ensuring compliance. And because Azure Cognitive Search includes prebuilt skills for language detection and translation, global companies can keep up with the regulations from all countries around the world, in many languages.  

Combining the strengths of two technology leaders

PwC is a member of the Microsoft One Commercial Partner program, an initiative that connects firms that are Microsoft Partner Network members with its technologies and expertise, along with joint selling opportunities. Sharma says, “Our relationship with Microsoft is critical to advancing the technology that we continue to develop.” 

Teaming up with Microsoft to create the PwC ROI solution is a triple win—for PwC clients, PwC, and Microsoft, according to Sharma. PwC partners with many cloud providers. Microsoft offers a globally available cloud in Azure in addition to maturity in AI. “Our purpose statement is to build trust in society and solve important problems,” she says. “As part of this commitment, we want to continue developing AI solutions and provide ease of use to our customers. Microsoft shares our passion for helping to make things easier for individuals and companies.”

Find out more about PwC on Twitter, Facebook, and LinkedIn.