Skip to main content

AI - Beyond a buzzword in enterprise IT

(Image credit: Image Credit: Geralt / Pixabay)

Companies are constantly throwing around the term “AI” as a trendy buzzword. So much so, that the true meaning has become lost somewhere along the way. As leaders grapple with a fast-changing environment where productivity, efficiency and reliability are all business-critical, it’s important to understand what tools are available and how to best use them.

I believe it all boils down to just a few components of AI - how those components different from other technologies out there, like RPA, and how to uplevel AI to reach higher levels of automation.

What is AI, really?

In contrast to the three types of AI, it’s important to understand that there are four main components of AI, which can be leveraged for different use cases.

Structured vs Unstructured Data

Everything starts and finishes with data. Enterprises rely on accurate data for every decision from planning to evaluating to predicting.  The challenge lies in the massive amount of data collected as well as the various formats in which the data is processed. We’ve been dealing with structured data for decades, which is easy enough to search, collect and analyze because it follows a predefined format. Typical RPA can sufficiently handle large amounts of this type of data. Unstructured data, on the other hand, can cause massive headaches, as there is no easy way to sort and extrapolate meaningful data to derive actionable insights. In many cases, companies cannot avoid manual processes to sort this type of data before running it through software systems. This is where AI is a truly valuable asset, because it can sort through massive amounts of unstructured data and understand how to process regardless of format. I’ve found that the two terms - RPA and AI - used interchangeably, but there is a big difference. RPA can do things that humans tend to do mechanically, only faster, while AI can do things humans cannot do (or cannot do at scale).

Semantic understanding

The second part of AI is semantic understanding, which is the ability to go through unstructured data and learn how, where, and why it is relayed—and guide what to do with it. AI can read and understand data like a human would. For example, if I work in the Finance department of a large organization, I may need to process hundreds or even thousands of invoices, expense reports, and other related pieces of information coming across my desk. Where RPA and OCR can read the documents and pull data if the information is exactly where it should be on a template, AI takes the next step to understand line items on receipts or match POs to invoices submitted to streamline the process. It understands the data and knows where to put it for processing. Using models, it can also match to company policies or regulatory compliance requirements. It goes beyond cutting and pasting information to truly understanding the data and being able to understand how and where the information fits into other processes.


Another differentiator of AI is that it can make informed, logical decisions based on its ability to understand and learn from data. For example, in manufacturing, there are a number of considerations that go into the go/no-go decision for product development. This process is traditionally handled manually - with one person, or a team, digging through a pile of historic data around purchasing, complaints, recalls and sales. By applying AI, this decision can be computed automatically, and with confidence, removing human error and eliminating hours of manual work.

Learning and Adapting

Now this is where we get into the biggest and, for some, most important benefit from using an AI product or platform. There are other technologies that, when combined, produce similar outcomes in the above examples. However, what makes AI so unique and valuable is that while going through the three steps above, the system has also been learning. It remembers that your team went back and approved certain items or which decisions were accepted or rejected - and it adjusts accordingly. This is seen in a lot of our daily lives when we engage with e-commerce companies like Amazon. The AI isn't just collecting our information to be able to provide a report on our shopped items and spend. Instead, it’s feeding Alexa so she can provide recommendations based on these previous decisions. In the beginning, these recommendations may feel a little off, but as we continue to provide feedback - many times through purchase behaviors - the recommendations get more and more accurate as the AI better understands us.

Self-driving AI

Now that we understand what AI is, another area with blurred lines is in the application of the technology on manual processes. We often see the terms “automated” and “autonomous,” and it can be difficult to understand the differences. The distinction lies in the ability to take an action as opposed to make a recommendation, as we saw above.

One of the best analogies I use is that of self-driving cars. Nowadays, we have a ton of great automated features, including alerts that beep if there are cars in your blind spot, flashing lights if you are veering into another lane or even guidance for parallel parking. These are all great improvements, but it is still required that the driver is 100 percent present and aware. In order to be autonomous, the car would be smart enough to take into account all of that data and actually drive on its own - without humans.

With business, automation streamlines some processes by doing the work normally done by a team - extracting data from thousands of documents, reports, invoices, etc. - but a truly autonomous scenario would be if it could then classify, analyze and make decisions, with confidence. It all sounds like a future dream, but that future is now. By truly understanding what real AI is and how to leverage it for your business, you increase not only efficiency but also output as you build and scale.

Kunal Verma, CTO and Co-Founder, AppZen (opens in new tab)

Kunal Verma is CTO and Co-Founder of AppZen, the disruptive leader in Finance AI. He is a published author with over 50 refereed papers and holds several granted patents.