It’s been a while since we launched Above Trends, Beyond Insights. What began as an experiment in editorial positioning has gradually become a method: looking past headlines to understand how technology actually behaves inside organisations.

When AI becomes the buyer: who is accountable?

AI agents for ecommerce are moving beyond recommendation. They can interpret a goal, compare options, apply constraints and select a product. In some cases, they can also complete the purchase or get very close to it. What used to support decisions is starting to execute them.

Payment networks and infrastructure providers are already building systems where agents can operate within predefined rules. Transactions can include authentication layers, spending limits and approval mechanisms. The direction is clear: agents are being integrated into real purchase flows, not just experimental interfaces.

The real change is about control.

When an agent selects a product, evaluates a price and chooses a supplier, the decision is no longer fully human. At that point, responsibility becomes harder to assign.

If the agent buys the wrong product, at the wrong price, or from a non-compliant vendor, the issue is not technical. It affects contracts, procurement policies and, in some cases, legal accountability. The question is simple: who is responsible for that decision?

There is no stable answer yet. Responsibility can sit across multiple layers: the user who activated the agent, the company defining the rules, the provider of the model, the merchant, or the payment infrastructure enabling the transaction. This fragmentation is exactly why current developments focus on guardrails.

Limits, approval flows, traceability and auditability are becoming essential. Without them, delegation introduces risk that companies cannot manage.

There is also a second layer that is often overlooked. The outcome depends on how the task is defined. Preferences, constraints and priorities shape the agent’s behaviour. In practice, the quality of a purchase depends on the quality of the instruction behind it.

Until now, this has been mostly an internal issue for companies using AI systems. With agentic commerce, it extends to consumers. Not everyone will define goals and limits with the same level of precision. As a result, two users can obtain very different outcomes using the same technology.

For companies selling to consumers, however, the challenge is quite immediate. Products will no longer be interpreted only by people, but also by agents.

This means rethinking how catalogues, product data and offers are structured and described. Information needs to be precise, consistent and machine-readable. Availability, constraints and conditions must be unambiguous.

In this context, visibility will depend less on how a product is presented to a user, and more on how it can be interpreted and selected by an agent.

TREND TRACKER

From recommendation to execution: how agentic commerce is emerging

Ecommerce platforms have spent years refining recommendation systems and personalisation. At the same time, conversational AI has changed how users search and interact with information. In parallel, payment infrastructures have become more modular and API-driven.

The combination of these elements is now enabling a new layer, where an agent can move from discovery to decision and, increasingly, to transaction.

Looking ahead, four developments appear likely.

First, agents will become more specialised. Rather than general-purpose assistants, they will be optimised for specific categories, recurring purchases and predefined policies.

Second, data structure will become a competitive factor. Products, availability, pricing and conditions will need to be exposed in a way that agents can interpret reliably.

Third, interoperability will matter more than interfaces. Without shared protocols, agentic commerce remains fragmented and difficult to scale.

Fourth, governance will become a differentiator. The ability to define rules, enforce limits and ensure accountability will be as important as the underlying technology.

Want to learn more?

• The agentic commerce opportunity - McKinsey

• Intelligent Commerce - Visa

• Agent Pay - Mastercard

• Retail Agentic Commerce Blueprint - NVIDIA

QUICK INSIGHT

Staying informed about AI without losing perspective

Keeping up with AI is a matter of filtering better.

1. Choose a small set of reliable sources
Follow a limited number of high-quality sources, and stay consistent. Mixing institutional reports, strategic analysis and market signals is usually enough to maintain a clear view.

2. Separate demos from real adoption
Many technologies look mature when presented. Fewer are stable inside real processes. Always ask where the solution is actually being used, and under what constraints.

3. Look at what is being built around the technology
Integrations, APIs, documentation, standards and compliance layers are stronger signals than announcements. They show whether a trend is becoming operational.

4. Focus on use cases, not capabilities
Capabilities evolve quickly and are often overestimated. Use cases reveal where value is already being created and where limits still exist.

5. Keep a simple evaluation framework
Ask three questions: what problem does this solve, for whom is it already working, what is still missing. This helps keep analysis grounded and avoids overreaction.

6. Accept that your view will need to change
AI evolves quickly. Staying realistic means updating your perspective as new evidence emerges, without committing too early to strong positions.

Following AI effectively today means staying informed without being reactive. Clarity comes from selection, not volume.

WHO IS BABINI MAZZARI

Our Value Proposition

Babini Mazzari is the strategic IT partner for European companies looking to navigate digital transformation in a structured, pragmatic, and sustainable way.
We don’t just deliver technical solutions - we work as an extension of your internal team, helping you integrate systems, optimize processes, and lead change with clarity and competence.

Our approach is built on listening, transparency, and a strong results-driven culture. Whether you're scaling, modernizing, or rethinking your operating model, we support every client with the right tools, clear methodology, and long-term vision.
Above Technology. Beyond Solutions.

📎 [Our services]
📎 [Our Manifesto]