insight magazine

Don’t Be Out-Teched

You don’t have to be one of the Big Four to deploy the latest tech tools in your firm. By Ryan Watson, CPA | Fall 2017


Much has been made about artificial intelligence (AI) in the media, both for its promise of efficiency and also for the risks it poses to a variety of industries, including accounting.

Just this past summer, our article, “Accounting for AI,” highlighted IBM Watson’s big debut into our space and the near-term implications it will have on our profession at large. That dialogue, though, at least so far, has primarily centered on enterprise executions of AI, like KPMG and H&R Block leveraging Watson to deploy custom AI solutions to their business challenges. And while that’s exciting for teams at those firms, what’s a small or regional accounting, tax, or advisory firm to do without the resources of in-house developers and data scientists?

Fortunately, as AI becomes mainstream, the opportunities to leverage it to automate work and deliver better client service are increasing daily. In fact, it’s likely your firm is already benefiting from AI and machine learning.

Here are three AI and machine learning technologies any size firm can start using right now.

Image Recognition

Image recognition is being used by firms of all sizes to automate everything from tax workflows to accounts payable and receivable—sometimes without them even realizing it. Popular apps like Hubdoc, Receipt Bank, and Expensify are all examples off-the-shelf web and mobile apps that are saving accountants and their clients from hours of manually processing transactions and expenses, but did you know these are all powered by image recognition and machine learning algorithms?

Financial document capture apps like the ones mentioned use an optical character recognition (OCR) process to scan paper bills, files, receipts, and statements, read the data, and interpret and digitize it. Using expense reporting as an example, data entry is as simple as having your client snap a photo of a receipt with their smartphone and the information is automatically extracted and populated in the client’s accounting platform, like QuickBooks or Xero. OCR is already employed in many facets of our everyday lives, including the tollbooths many of us pass through each day— OCR technology is what’s behind the automatic license plate recognition system that charges our I-Pass accounts when we forget our transponders at home.

Hubdoc employs an extra layer of human quality assurance to certify the correct information is extracted by its technology, but it still means no data entry or filing for you or your client. What’s exciting about this, besides the time-savings and convenience, is that we’re really just getting started on our AI journey.

“Leveraging mature technologies, like OCR, and some manual quality assurance, we’ve been able to streamline the bill and receipt capture process for our customers. But this just scratches the surface of what we think we’re capable of,” says Hubdoc Co- CEO Jamie Shulman. “As we start to truly take advantage of AI and machine learning, our goal is to make the end-to-end expense management and bill payment process truly automated and instantaneous.”

This is a technology that’s easy to tip-toe into. You can start with automating a single workflow, like accounts payable. Working within the comforts of your clients, select the right tool, perhaps Hubdoc, and integrate it with a digital general ledger and, if applicable, a bill payments suite. From here, Hubdoc can sync with vendor portals to automatically ingest all bills for payment or receive them through email or standard uploads. The image recognition software will automatically identify the date, vendor, amount, and after some training by the firm and client, the account code. It’s really easier than you’d think.

Recommendation Engines

Receiving recommendations from some form of AI may sound hyper-futuristic, but when you consider the following examples, you’ll see there’s a common thread here: AI and machine learning systems are already part of our personal and professional lives.

When Netflix recommends what you should watch next, it’s based on algorithms that pull from the taste graph of their entire global subscriber base, taking into consideration what’s available by region and similarities between content and user behavior—this is a recommendation engine driven by machine learning.

Similarly, accounting software companies are implementing machine learning systems within their platforms to help prevent and correct user mistakes. Take cloud-based accounting software provider Xero for instance; coming off of an intensive, year-long project where developers looked at the mental processes users go through, they’ve built a machine learning model that learns user invoice coding behaviors and mistakes, and then notes the corrections accountants and bookkeepers make in order to automatically offer recommendations and corrections as users work within the system in the future.

Xero’s machine learning system, already available to users, improves with use, meaning it “learns” as you’re feeding it data. And have you ever noticed how your smartphone tries to predict the next word of your text message, or how Google seems to know the search term you’re trying to put it? The point is, you’re already benefiting from recommendation engines, and they’re only going to become better—and more prevalent. The more you seek out the tools deploying this technology, the easier your tasks will become.

Smart Assistants

Do we have any Iron Man fans here? Have you ever imagined having your very own J.A.R.V.I.S. to organize your life? Apple’s Siri and Amazon’s Alexa are starts, but full-fledged AI personal assistants may be closer than you think. While smart assistants aren’t quite capable of toasting your bread yet, they can organize the myriad of appointments and requests for meetings that pile into your inbox.

Smart assistants use natural language processing (NLP), a technology that enables software to understand the actions and language of humans. NLP gives smart assistants the ability to not only interpret the language but to understand the intent behind it. Smart assistants can also learn the unique preferences of the user, such as their preferences for meeting length or location.

Amy Ingram, for example, is a smart assistant developed by, or an AI personal assistant who schedules meetings on your behalf in an intelligent and human-like manner. The beauty of Amy is that it’s easily integrated into your workflow by CC’ing “her” into your emails. Amy then reads the emails and communicates directly with the contact on your behalf, arranging meetings and appointments based on your personal schedule and preferences of times, locations, and meeting types. The most significant benefits of Amy are her efficiency, human-like capabilities, and ability to respond in a thoughtful and intelligent manner—people actually think she’s a real person. Talk to her like a real-life assistant and the magic of her AI takes over.

Big Data Player

We don’t have to tell you that your business deals with a lot of data. But businesses today do need to be reminded that this data is only useful if it’s monitored and leveraged for strategic decision-making and effective risk management.

The truth is that you’re already a big data user; you just need to embrace it. If you’re a user of Facebook, Google, payment processors Stripe or PayPal, or any of the other platforms mentioned in this article, for instance, you’re contributing to our evolving big data ecosystem, helping to plant the seeds of future AI advancement. This data serves as the foundation upon which AI and machine learning models are built, refined, and reinvented.

For example, Facebook utilizes insights from its big data to provide targeted advertising based on a consumer’s preferences. PayPal leverages data from its closed-loop network to improve customer experiences and detect and prevent payment fraud. And Stripe utilizes data to analyze and identify patterns of suspicious activity in payments on its network to safeguard your transactions. Hubdoc, Xero, and Google are all building big data architectures that will automate more of our business workflows in the future.

The point is that you don’t have to be one of the Big Four to take advantage of big technologies. AI, machine learning, and big data are all around us. The earlier we all embrace change and begin employing the tech tools available to us, the better we’ll be positioned to be firms of the future.

Ryan Watson, CPA is a founder and principal of tech-savvy accounting firm Upsourced Accounting and Xero’s Midwest ambassador.

Leave a comment