EMA Journal • Fall 2025 • Store Equipment

From automated transaction processing to predictive analytics, artificial intelligence (AI) is reshaping how energy marketers operate, compete and connect with customers. The question is no longer about whether AI will impact your business — it’s all about how quickly you can act on the potential of AI while navigating its challenges.

To dive deeper into the real-world use cases of AI, we recently spoke with Justin Baxley, SVP of product management, retail, at PDI Technologies He shared his insights on how AI is serving as a collaboration tool to redefine problem solving and drive practical outcomes for energy marketers across their daily operations.

Q: What’s the competitive upside of adopting AI now — or the downside of not using it?

A: With the pace of AI innovation continuing to accelerate, the competitive gap between successful AI adopters and non-adopters is also widening rapidly. And it really puts the pressure on businesses to at least try it. For example, AI can help you quickly streamline customer service, optimize your pricing and create more targeted marketing campaigns. You can also reduce your overhead costs while improving accuracy for forecasting and inventory management. If AI isn’t part of your tech stack, you’re going to find yourself at a competitive disadvantage.

A: We’ve seen the most success with companies that introduced AI solutions based on the business outcomes they wanted to achieve rather than what the technical capabilities might deliver. So, it’s just like any other technology decision — you need to identify your goals and look for specific use cases or opportunities. Start by building a strategic AI roadmap for high-impact or low-risk applications. Get some wins under your belt, then keep expanding. A methodical, phased approach tends to work better for long-term success. Think of AI as a way to augment your team’s capabilities — and give them the right training to thrive in an AI-enhanced work environment. It’s also important to make sure you come in with the right expectations about ROI. There are massive breakthroughs happening, and that can get a little theatrical. Make sure your excitement curve doesn’t exceed the math of reality.

Q: Do you need a certain type of infrastructure to implement AI effectively?

A: Most people aren’t spinning up new AI implementations from scratch with cutting-edge researchers and a bespoke tech stack. The super-majority of the most accessible AI applications will be cloud-hosted and in many cases embedded in one of your existing providers’ infrastructures. So, infrastructure challenges are more often tied to data. AI isn’t psychic — it’s still technology with inputs and outputs, no matter how rich the data has become. That means it needs access to meaningful data sources. If you plan to create your own tools, you’ll need a robust data strategy, likely tied to an existing cloud provider and enabled by tools that play well in that cloud services environment. But don’t underestimate the challenge of data preparation and governance.

A: The recent breakthroughs in large language models and Gen AI are machine learning models at their core. But they’re often broad or deep in ways that aren’t necessary for every application. More established machine learning algorithms are comparatively cost-effective and, in many cases, more fit for purpose. So, in predictive use cases like pricing, inventory optimization or demand forecasting, don’t turn your nose up at hardened machine learning solutions.

Q: Can AI help with anticipating shifting consumer behaviors or market trends?

A: Yes, the same capabilities extend into proactive decision making around forecasts. You can leverage historical data to predict when customers are most likely to increase their fuel purchases or act on a specific offer or promotion. When it comes to market trends, predictive models can help you forecast fuel demand fluctuations and predict supply chain disruptions before they happen. In all of these cases you need a solid foundation of high-quality data.

A: Intelligent automation and real-time decision making are already driving advancements in transaction processing — especially for areas like payments and fraud detection. The payoff would be fewer processing errors, faster reconciliation and the ability to handle peak volumes. The goal is to streamline transaction processing while improving security and accuracy while minimizing human intervention.

Q: Speaking of communications, does AI support any strong documentation upgrade opportunities?

A: AI is a real time-saver when you want to streamline document creation or implement new quality control standards, especially around regulatory compliance issues. It’s much easier to generate contracts, compliance reports and customer communications at scale. That means less administration overhead, faster responses to RFPs, and greater consistency across the board. AI can automatically customize documents for specific customer segments or regulatory requirements.

Q: That brings up some questions about regulatory compliance. What are some of the legal considerations with using AI?

A: First, you must have clear policies and review processes for AI-generated content usage, IP protection and general usage guidelines. Your legal and technology teams have to work in lockstep to develop guidelines, train employees and ensure the necessary safeguards to protect the business. It’s critical to understand the usage and data policies of the tools you interact with and also regulate how and where your employees use AI — along with where your data goes.

A: I’d just say it’s clear that AI offers unprecedented opportunities to improve efficiency and stimulate growth. However, there are three important things to remember. First, AI projects often fail because of weak data strategy or the pursuit of the gimmick instead of a clear business value. You have to get those things right. Second, the most recent developments are in many ways still experimental. You have to manage AI agents just like employees for their behavior and content. Finally, AI isn’t a total replacement for human experience and judgment. You’ll always need to build the human connections that make up a successful energy business.

Justin Baxley is SVP of product management, retail, at PDI Technologies, EMA Platinum Corporate Partner. You can connect with him on LinkedIn.