Defining Transactional AI
Over the last several years, the AI industry has focused heavily on analytical AI, generative AI, and more recently agentic AI. While each of these categories describes important capabilities, I observed a disconnect between how AI was being discussed and where some of the most valuable enterprise AI opportunities actually existed.
The concept of Transactional AI emerged from a discussion with Andrew Mead, VP Worldwide Sales IBM Z Application & Platform Software, regarding the role of AI within enterprise transaction processing systems. The observation resonated with me because it aligned with a pattern I had been seeing across client engagements. While predictive AI, machine learning, deep learning, and real-time inferencing technologies had existed for many years, organizations lacked a clear way to describe a distinct class of enterprise AI where intelligence is applied directly within transaction processing systems to improve business decisions at the moment they occur.
The more I explored this pattern, the more convinced I became that it represented a distinct class of enterprise AI. Transactional AI emerged as a way of defining and communicating that distinction. The objective was not to introduce a new technology. Predictive AI already existed. Machine learning already existed. Deep learning already existed. Real-time inferencing was already occurring within enterprise systems. What was missing was a way to describe this emerging category of AI and articulate its strategic importance to organizations whose most critical business decisions occur inside transaction processing systems.
The idea was built on a simple observation. Organizations already make millions of decisions every day. Payments are approved or declined. Claims are adjudicated. Transactions are monitored. Customers are onboarded. Credit applications are assessed. Service requests are routed. These decisions already exist and are often executed at enormous scale.
The opportunity presented by AI is not to create new decisions. The opportunity is to improve decisions that already exist.
This distinction fundamentally changes how organizations think about AI. Rather than viewing AI as a separate capability operating alongside the business, Transactional AI views intelligence as an extension of operational decision-making. The objective is not simply to generate more insights. The objective is to improve business outcomes by enhancing the quality of decisions at the point they occur.
At the time, much of the industry’s attention was focused on model development, AI infrastructure, and increasingly sophisticated AI capabilities. While these areas remain important, they represent only part of the opportunity. Many organizations were already applying intelligence within payments, fraud detection, claims processing, customer onboarding, lending, and other mission-critical processes. Yet these initiatives were often discussed as isolated use cases rather than as manifestations of a broader pattern.
Transactional AI provided a way to define and communicate that pattern. Rather than starting with technology or predefined use cases, it encouraged organizations to focus on the decisions embedded within business processes and the outcomes those decisions influence. This shift in perspective often revealed opportunities that remained invisible through traditional AI discovery approaches because the discussion moved from applications and systems to operational decision-making itself.
One of the most important observations behind Transactional AI is that not all decisions are equal. Some decisions have little impact on business performance. Others directly influence revenue growth, operational efficiency, fraud losses, customer satisfaction, regulatory compliance, and risk exposure. The value of AI is often determined less by the sophistication of the model and more by the importance of the decision being improved.
This realization connected naturally with my broader work in Business-First AI Strategy and AI Economics. Rather than starting with technology capabilities, the discussion begins with business outcomes. Which decisions influence those outcomes? What economic value is associated with improving them? How can intelligence be applied in a way that creates measurable business impact? These questions frequently lead organizations toward opportunities embedded within transaction processing systems because that is where many of their most important decisions already occur.
As I continued to develop and apply the concept, I became increasingly convinced that some of the highest-value AI opportunities exist within the operational core of the enterprise. These opportunities are not defined by the size of the model, the volume of data, or the scale of the infrastructure. They are defined by the importance of the business decision being improved. Whether the decision involves approving a payment, detecting fraud, adjudicating a claim, onboarding a customer, assessing credit risk, or routing a service request, the underlying principle remains the same: better decisions produce better business outcomes.
Today, I continue to use the concept of Transactional AI to help organizations identify, prioritize, and operationalize high-value AI opportunities. Whether the objective is reducing fraud losses, improving payment authorization rates, accelerating claims processing, strengthening customer onboarding, improving customer retention, or increasing operational efficiency, the focus remains on understanding where critical decisions occur and how intelligence can improve them.
I believe Transactional AI represents one of the most important and least understood classes of enterprise AI. Not because the underlying technologies are new, but because organizations continue to underestimate the value that can be created by improving the operational decisions they already make every second of every day. As AI continues to evolve, the organizations that achieve the greatest business impact will not necessarily be those with the largest models or the most sophisticated infrastructure. They will be the organizations that successfully apply intelligence to the decisions that matter most.
