The first time I interacted with LLMs a couple of years ago, I couldn’t initially decide whether this would be another gimmicky fad or a truly life-changing piece of technology. Artificial Intelligence has been around for a long time and already extensively used in operational tasks to drive automation, data analysis and classification. However, what has fundamentally changed is the concept of Generative AI. AI systems now have the ability to generate new content, new data, new answers, which carry high impact on businesses, consumers, industries, and more broadly on management systems and decision-making processes. Generative AI is now capable of proposing an infinite number of right answers and the challenge we, as leaders, face today is how do we choose? How do we evaluate these models?
The Generative AI value gap, to quote C. Kozyrkov can be explained through 6 major points:
- The importance of clarifying who gets to call the ‘shots’ when there are virtually unlimited right answers proposed by AI systems.
- The importance of clarifying what organisational strategies are and how ROI should be measured.
- Be the author of meaning. That is for leaders to be clear about their thoughts, and be clear and precise in their communication approach when managing technical subjects.
- Thinking in terms of ‘good enough’. Have a clear definition of what good enough looks like and stay truthful to it.
- Leverage a ‘Human in the Loop’ approach where human readings can evaluate outputs according to wider strategic goals.
- Leverage randomised control tests to evaluate results scientifically.
- Tie the output back to the business. Why would this be an improvement? How would it affect business performance?
In my fifteen years of experience in Digital and eCommerce, I've witnessed numerous innovations that promise to revolutionize how we buy and sell online. From mobile commerce to voice shopping, to the rise of marketplaces, each wave brought incremental changes to the digital retail landscape. However, few developments have the potential for such profound transformation as what I’m now seeing with agentic AI. New shopping assistants living on new interfaces ready to generate the perfect solution that responds to online shoppers’ needs.
We find ourselves at a fascinating inflection point. The conversational AI interfaces that dominated headlines in recent years are evolving into something potentially more powerful and autonomous fueled by exciting generative capabilities.
In the year 2024, the market value of agentic artificial intelligence (AI) stood at 5.1 billion U.S. dollars. It is anticipated that this market value will surpass 47 billion U.S. dollars, with a compound annual growth rate of over 44 percent, as reported by Capgemini.
According to McKinsey's "The State of AI in 2023" report, 40% of respondents in retail and packaged consumer goods have already adopted AI for product development and service optimization, demonstrating the accelerating pace of AI adoption in commerce (McKinsey, 2023). Behind these statistics are real businesses racing to understand and implement AI capabilities that go beyond simple automation.
In this post, I'll explore how agentic AI systems could reshape the commerce landscape, examine the frameworks that define their capabilities, and analyze how businesses are beginning to explore these technologies. Most importantly, I'll consider the critical question: Are we witnessing the early stages of a fundamental shift in how eCommerce functions?
Looking Ahead: The Evolution of eCommerce
The evolution of AI in commerce—from basic recommendation engines to the brink of autonomous agentic systems—signals a profound shift in how businesses and consumers interact. While truly agentic commerce remains more vision than reality today, the foundational technologies are rapidly advancing across multiple domains.
The question is no longer if AI will transform commerce—it already has—but rather how quickly and to what extent agentic capabilities will mature, and what new forms of commerce they might unlock.
For businesses, this calls for a strategic approach that balances current limitations with future readiness. Practical steps might include:
- Improving data readiness: Structuring product data, pricing, and inventory in machine-readable formats. As one eCommerce platform architect put it, “The companies that organize their data with AI readability in mind today will have a significant advantage tomorrow.”
- Experimenting with automation: Deploying existing AI tools in targeted areas like inventory optimization or personalized recommendations. These pilots build internal capability and foster customer trust in AI experiences.
- Tracking technological developments: Staying up to date on advances in multimodal AI, planning algorithms, and autonomous agents that are rapidly pushing the boundaries of what’s possible.
The path toward agentic commerce is likely to be evolutionary, not revolutionary—with human oversight continuing to play a vital role. In my own testing of emerging shopping technologies, I’ve found that the most effective systems are those that thoughtfully integrate AI with human judgment, leveraging the strengths of both.
What excites me most isn’t just the potential for greater efficiency, but the opportunity for more intentional consumption. By delegating routine decisions to AI, we may create space for more mindful, value-driven choices.
As we move forward, one thing is clear: the relationship between consumers, brands, and the intelligent systems connecting them will continue to evolve—bringing both opportunities and responsibilities. By staying informed and adaptive, we can help shape a future of commerce that is not only smarter but more meaningful.