A practitioner's guide to AI in the enterprise

Here at Vest Side Stories, my intent is to offer a wide variety of expertise and information intended to help people build better businesses. Whether you’re a first-time founder or a serial entrepreneur, a C-level executive or an up-and-coming leader, whether you work at a Fortune 500 company or a 10-person start-up, this blog — hopefully — contains something of value for you. To that end, I’ve spent much of the past year talking to industry experts and luminaries, diving into the specifics of their respective realms to find guidance and wisdom that I simply cannot provide alone.

But as summer gives way to fall, I feel inspired by the changing seasons to follow suit. It has been a long while since we dove headfirst into a multi-part series here on VSS, and today we’re standing on the precipice of what will most likely be the greatest technological revolution of our lifetimes. I’m talking, of course, about the proliferation of mainstream Artificial Intelligence.?

The conversation around AI is a fascinating one that is in danger of quickly becoming an echo chamber, one that provides inspiration without value or instruction. The conversation around AI is still largely hypothetical — “what will it do for us, and won’t that be amazing?” Well, sure. But what does that conversation actually accomplish, other than to perhaps birth a new generation of science fiction writers? As an industry, it is no longer our job to discuss AI in the abstract. It’s our job to turn AI into something tangible, useful and transformative. That undertaking starts with understanding how AI will impact specific use cases in the enterprise, the daily functions that every business must have in place to succeed, to innovate, and to bring all of what is promised to fruition. Essentially, the first step in bringing this technology to the mainstream is to learn, as an industry, how to utilize it most effectively ourselves.?

In this series, we will discuss a wide spectrum of enterprise use cases, from sales and marketing to IT and engineering to supply chain and procurement. I want to help you decipher the value through the noise, because even though this is an emerging space there are right ways to approach these solutions and a wrong way. And while AI is incredibly powerful, today its value is still largely dependent on the quality of human touch involved. If the tools aren’t used the right way, the results we expect from transformative tech simply won’t be there.?

We must be conscious both about how we build and buy these products. It is important to have a clear set of objectives in mind, which comes from a clear understanding of the specific use cases we’re trying to address. Consider this a framework for use cases in the enterprise where I see AI as a major opportunity. Let’s start by laying out some core objectives for both builders and buyers:

1. Does the technology, product or solution in question solve a clearly defined business problem? Making something faster, for example, doesn’t necessarily make for better outcomes. It’s possible to do quite a bit fast, and completely wrong. This can cause insurmountable problems downstream. Be sure you’ve clearly defined the business problem, and can articulate how AI can specifically solve for that problem.?
2. Do you (or your customers) have the infrastructure in place required to benefit from AI? There is exponentially more data available to us today than ever before, which is the fuel that makes AI run. But the ability to ingest, transport and process that data at scale and speed is a prerequisite. On top of that, there sits the intelligence layer — the algorithms that have allowed for LLMs to come into maturity. All of these things are now possible, but the infrastructural requirements for making them all work together must be clearly understood.?
3. Where are you going to move the needle most? A good use case for AI is one with lots of available data, and data that will grow exponentially over time. But it’s also one that has existed and will continue to exist through times of societal and technological transformation. Take for example, fraud detection for credit cards. The use case is clear, with millions of dollars in fraud each year. It is data-intensive, and that data continues to grow. We can create intelligence around it that will learn and improve over time. These criteria must then be applied to each use case to determine whether or not AI is the right fit. For some, it won’t be — and that’s okay. If you’re not getting the intended results, AI for the sake of AI has very little point.?
4. Can you clearly and easily measure output, thereby allowing you to adjust your strategy on the fly as-needed? Let’s refer back to the example of credit card fraud detection. Any AI built to identify credit card fraud will need to be perpetually trained on new transaction types in order to remain effective.?
5. Is this a use case that is primed to be impacted significantly by future legislation? Data enforcement and regulation is already starting to expand. Regulations around a technology like AI tend to roll out in one industry or vertical at a time, as the technology is adopted and popularized within that industry or vertical. It is critical to understand where and when those regulations will impact your business, and to ensure you’re not set up for regulatory issues concerning your data, how it was collected or how it is managed and protected down the road.?

Through the course of this series, we’ll explore each of these parameters in more detail as we dig into the most common enterprise use cases and how AI will impact them. As we go, I want to hear from you — which cases do you find most compelling? What are your own personal experiences with deploying AI in the enterprise, both successes and failures? We are ushering in a new era, one that will only fulfill its promise if we take a collaborative approach to the journey. So let’s talk.