Justin Thomas, VP Sales, EMEA North, Akeneo explains how product data separates the winners from the losers in AI commerce.
AI has transformed the role of product information management (PIM). Once defined as a series of mechanical processes (attributes, gaps, descriptions, enrichment) in the service of completeness and speedy onboarding, PIM responded to the traditional search journey.
Now, AI has rapidly redrawn the boundaries of how products are discovered, evaluated and purchased, which requires a different approach to data management. As things stand, AI has split brands and retailers into two different camps: those that simply manage product data, and those that can execute using it.
In today’s eCommerce landscape, shaped by AI-driven agents, disintermediation through social platforms and marketplaces, increasing regulations, and API-first ecosystems, simple data enrichment is no longer enough to create a competitive advantage. Execution is the key and is defined by how product truth is governed, managed across channels and made ready for activation.
The most significant aspect of this shift is the rise of AI agents in the makeup of the buying interface. Embedded in search engines, marketplaces, social platforms and enterprise procurement tools, these systems go beyond scrolling through category pages or highlighting the brand narrative; they ingest product data, assess relevance, evaluate risk and make recommendations, often without ever touching a traditional product detail page.
It is important to understand that AI agents take everything literally. They depend on structured attributes, validated claims, consistent taxonomy and contextual signals to understand what a product is, who it is for, whether it complies with regulation and how confidently it can be recommended. If that data is incomplete, inconsistent or poorly governed, the agent doesn’t hang around, it moves on.
What this means is that the old measures of visibility through SEO are no longer of value. More important is the quality and reliability of underlying product data pipelines that feed AI decision-making.
Brands and retailers are also up against the fact that they are losing control over the channels they once relied on. Social commerce, marketplaces, retail media networks and AI-powered discovery tools increasingly sit between the consumer and the brand’s own digital experience. The traditional website is no longer the ultimate point of truth; it is just one of many activation surfaces.
This level of disintermediation puts a lot of pressure on product data platforms. Every channel, agent and algorithm now requires a different slice of product truth. Product descriptors can no longer be static; they must be governed and context-aware in order to be fully optimised for each decision moment. Governed product data gives AI agents the confidence to recommend, compare and transact, as opposed to ungoverned data which creates friction, hesitation or exclusion.
As a result, companies that treat PIM and PXM (Product Experience Management) as standalone tools for data clean-up before syndication quickly hit a ceiling, while those that embed them within a broader platform turn them into execution engines, able to dynamically orchestrate product narratives across ecosystems they no longer control.
Furthermore, once regulation is added to the picture, the room for error expands. From sustainability disclosures and digital product passports to safety, compliance and traceability requirements, AI systems are increasingly being asked not just what a product is, but whether its claims can be trusted. Unsupported assertions, outdated certificates or ambiguous data structures reduce conversion as well as heighten the risk of legal action and reputational damage.
The final piece of the puzzle is technical architecture. AI commerce thrives in agile environments, such as API-first ecosystems, where product truth can be accessed, recomposed and activated in real time. Static exports, manual workarounds and channel-specific hacks simply cannot keep up with the velocity of AI-driven decision-making.
This means that modern product data platforms should be interoperable, able to feed search engines, AI agents, marketplaces, retail systems and emerging channels from a single governed source of truth. Execution can then take place much more quickly, more confidently as it is reused and adapted more safely without breaking trust.
Product data platforms are effectively becoming the connective tissue of AI commerce itself, deciding where products appear, how they are understood, evaluated and chosen.
All of this elevates the role of IT and product data teams in ways many organisations have yet to fully appreciate. In an AI-mediated marketplace, performance is managed by data pipelines, conversion is shaped by governance, and visibility is enabled by interoperability. That makes (product) data teams the new gatekeepers of growth, fully in control of not just content but consumer trust.
These teams will enable their organisations to move beyond enrichment as an end goal and instead build execution-ready product truth that is governed, adaptable and built for machines and humans alike.
Share