AI should expand human capabilities, not eliminate human direction. Businesses must retain ownership of the data, logic and expertise that differentiate them.
Romain, your career has taken you from data science to strategy, operations, and revenue leadership at Dataiku, and now CEO of Akeneo. What early lesson still shapes how you turn technically promising innovation into lasting business value?
My advice would be to ride the wave early, focus on solving your customers’ problems and build strong partnerships with them. Open your network to them, involve them in shaping your roadmap and help their teams succeed. Success matters for both the organization and the people who trust you. That’s how customers become advocates who can carry your story further than any campaign. Shifting a market is an ultra-trail, not a sprint, and you don’t run it alone.
At Dataiku, you helped scale annual recurring revenue from $40 million to more than $250 million while leading global go-to-market operations. What changes when enterprise AI moves from an exciting technology to something customers depend on at scale?
The real test of enterprise AI isn’t whether it works in a demo, but whether businesses can rely on it in production. A successful pilot might work with a limited data set and clearly defined processes, but production introduces changing requirements, conflicting information and more complex workflows.
At Dataiku, a key part of scaling AI was enabling collaboration across disciplines. At Akeneo, the focus is on applying deep expertise in product data to make AI reliable in commerce. We understand the processes and people behind product data, which helps us see where an agentic workforce can deliver meaningful value in practice.
In commerce, product data is fundamental to making that happen. AI agents can’t make accurate recommendations when specifications are missing or if information conflicts across systems and marketplaces. That’s why we often say there’s no good AI without data. To achieve recommendations from AI agents businesses need to balance three priorities: velocity, comprehensiveness and trust. Product information needs to move quickly, meet increasingly specific customer needs and remain trustworthy as more decisions become automated.
As agentic AI moves software from generating recommendations to taking action, where do you see the most meaningful opportunities for enterprises today, and how should leaders decide which decisions are ready for greater autonomy?
The greatest opportunity with agentic AI lies in changing how people work. Today, a product data team might spend significant time manually filling database fields and maintaining information. With an agentic workforce, those same employees could instead direct AI agents to manage much of that work. We’re already seeing opportunities for specialized agents to support activities like supplier data ingestion, product enrichment, taxonomy mapping and translation.
I’d suggest a different perspective on keeping a human in the loop. I prefer to think about keeping humans at the helm. People should establish the direction, define the data structures and develop guardrails, while agents handle much of the execution.
That means defining clear boundaries on which actions agents can take independently and which require human review. Routine product attribute enrichment may require less oversight, while changes involving regulatory requirements, technical accuracy or brand claims may need closer review from product experts, who remain the authority on those decisions.
Also Read: AITech365 Interview with Kumar Chivukula, Chief Executive Officer at Opsera
Deloitte’s 2026 State of AI in the Enterprise report found that only one in five companies has a mature governance model for autonomous AI agents. What separates effective AI governance from policies that exist on paper but struggle to guide real-world decisions?
At Akeneo, we’ve approached this through the Akeneo AI Harness, which provides the permissions, business context, orchestration and traceability agents require to operate within established business rules. The objective isn’t to have employees review every individual action. It’s to allow agents to work efficiently while people retain authority over the decisions that matter.
This means defining what agents are permitted to access or modify, establishing approval workflows for sensitive decisions and ensuring their actions can be traced back to an accountable person. Without those safeguards, organizations risk falling into what we call an AI Slop Trap, where AI-generated content becomes increasingly difficult to manage, and from a consumer’s perspective, harder to trust. An agent with the wrong permissions, for example, could change thousands of product records or publish claims that haven’t been approved.
The bigger risk is losing visibility into who made a decision and why. As AI capabilities evolve, businesses need governance that can adapt without sacrificing control, allowing them to move quickly while maintaining confidence in the decisions their technology supports.
Poor data quality and fragmented systems can undermine even the most capable AI agents. Before scaling autonomous workflows across an enterprise, what foundations should businesses establish, and how can leaders tell whether their data is genuinely ready for AI to act on it?
A common mistake is assuming that because information exists somewhere on a website, AI agents can reliably use it. Information buried in large, unstructured fields can introduce ambiguity or cause agents to make inaccurate assumptions.
Data completeness and data readiness aren’t necessarily the same thing. A product can have every required field populated and still be missing information needed for a particular market, channel or use case. Businesses should establish clear readiness criteria and test whether their systems can identify authoritative information, handle conflicting inputs and support accurate decisions. At Akeneo, our Readiness Policies help businesses define and enforce those requirements across markets and channels.
We’re moving from a search-first digital experience to an intent-first one, making structured, contextual product information even more important. Standards such as ETIM and BMEcat, along with requirements like the Digital Product Passport, give product attributes a consistent structure and meaning. Businesses also need to continuously use signals from buyer questions, reviews and returns to identify gaps and improve the information their agents rely on.
Many organizations are adding AI to workflows designed around human decision-making. From your experience, what needs to change in how businesses assign responsibility, maintain human oversight, and redesign workflows so that AI can improve execution without creating new blind spots?
Organizations need to resist the temptation to simply automate the processes they already have. Automating a broken process only creates a faster broken process. Agentic AI creates an opportunity to rethink how responsibilities are distributed between people and technology. Employees who once performed repetitive data management tasks may instead coordinate specialized agents, monitor performance and improve the systems those agents use.
That requires moving beyond isolated AI applications. One agent might process supplier information, another enrich it and a third validate it before it reaches sales channels. All three work from consistent data and approval rules. As agents take on more of the execution, people need to focus on defining outcomes, which requires greater collaboration across IT, product and business teams.
One of the biggest blind spots is a go-to-market strategy still organized around individual channels, while product information sits with another team. Businesses need shared goals, clear ownership of product information and a consistent product story across channels.
At Akeneo, product information is becoming increasingly important to how products are discovered and evaluated through AI-powered commerce. As AI agents influence buying decisions, what must businesses get right about product data to ensure these experiences remain accurate, relevant, and trustworthy?
Your products now have two audiences: people and AI agents. People increasingly turn to AI first. An AI assistant might compare dozens of options and present only a handful to the shopper, so a product missing critical information may never make that shortlist. Product data used to support the customer experience once someone reached a website. Now, it can determine whether they reach it at all.
AI is closing the knowledge gap between consumers and experienced category managers, giving buyers the ability to ask increasingly detailed product questions. Every buyer is becoming an expert buyer, making comprehensive product information more important than ever. The market’s reflex is to generate more content, faster, but AI can’t replace product data expertise.
Recently, while I was researching a product, I came across a website that appeared entirely credible but raised questions about whether its content was genuine. Creating convincing fake websites and product reviews has become remarkably easy.
Earning trust will become one of the defining challenges in AI commerce. When product information differs between a brand’s website and a marketplace, AI struggles to determine which source is accurate. Brands need to anchor their information in verified, structured product data that they publish and govern themselves. As AI becomes more influential in product discovery, maintaining that consistency across channels will be essential to how products are represented and recommended.
In your thinking on PIM ROI, you introduced Net Emotional Footprint as a way to understand trust and user sentiment. How can businesses translate these signals into measurable outcomes, and what might they reveal about technology adoption that conventional ROI metrics overlook?
Businesses can begin measuring that transformation by looking at changes in manual effort, workflow cycle times, accuracy and the amount of time employees spend reviewing or correcting AI-generated work. These measures help product launches go to market faster, fewer errors and greater efficiency.
Net Emotional Footprint captures something those numbers can’t fully measure: how product information shapes confidence. For buyers, that shows up in the questions they ask, the reviews they leave and the reasons they return products. An organization might demonstrate significant time savings, but if employees are constantly correcting errors or questioning AI-generated outputs, the technology may not be delivering the value those efficiency metrics suggest.
I would encourage leaders to look beyond immediate efficiency gains and consider how AI and product information contribute to long-term business performance. ROI makes the case today, while Net Emotional Footprint helps build customer trust and brand value over time.
With Akeneo bringing product information and pricing intelligence closer together through its PricingHUB acquisition, how do you see these capabilities changing the decisions businesses can make, and what will it take to deliver that value through a coherent platform?
A price only exists in the context of a product. Every attribute, specification and point of differentiation helps explain what a product is worth, whether it can command a premium and how it compares with alternatives in the market. Yet product data and pricing often sit in separate teams and tools. The right product at the wrong price can still lose the sale.
Traditionally, businesses have used customer profiles as a proxy for buyer intent, but AI is making it possible to understand that intent more directly. Two people may both be shopping for a white T-shirt but have very different needs. Understanding the differences in their needs, their preferences, and importantly, their budgets, allows businesses to recommend more relevant products and connect pricing to the product itself rather than assumptions about the individual.
With PricingHUB, we’re bringing pricing intelligence together with product information as part of Akeneo’s Product Cloud strategy. This helps pricing, e-commerce and category teams work more collaboratively, respond faster to competitive conditions and make more informed pricing decisions. Ultimately, AI-driven commerce experiences gain the context to recommend the right product at the right price.
Having experienced enterprise AI scaling at Dataiku and now leading Akeneo as agentic capabilities expand, what principle do you believe businesses must preserve as they give AI more responsibility, and what will distinguish those that scale it successfully from those that struggle?
As AI agents take on more of the execution, leaders need to focus on managing outcomes rather than every individual step. That means defining what success looks like and giving teams the flexibility to achieve it.
AI should expand human capabilities, not eliminate human direction. Businesses must retain ownership of the data, logic and expertise that differentiate them. Deploying more agents is the easy part. The greater challenge is rethinking how businesses bring products to market, with product data at the center.
As technology becomes more autonomous, people must still establish boundaries and remain accountable for outcomes. That’s why I keep coming back to humans at the helm. AI may do more of the work, but people remain responsible for where it takes us.
Thanks, Romain!
About Romain Fouache
Romain Fouache is the Chief Executive Officer of Akeneo, bringing more than 20 years of experience in scaling high-growth B2B technology companies. Before joining Akeneo, he served as Chief Operating Officer and Chief Revenue Officer at Dataiku, where he helped drive business growth and expand the company’s enterprise AI capabilities. A graduate of École Centrale Paris, he also holds an MBA from NYU Stern. At Akeneo, he focuses on driving global growth, product innovation, and AI-powered product experiences for brands and retailers worldwide.
About Akeneo
Akeneo is a global Product Experience (PX) company providing an agentic-first Product Cloud platform that helps brands, manufacturers, distributors, and retailers manage, enrich, and optimize product information. Its platform centralizes and orchestrates product data, pricing, and digital assets to deliver consistent and engaging product experiences across e-commerce, mobile, social media, marketplaces, and other sales channels. By combining product information management, AI-powered capabilities, and intelligent automation, Akeneo helps businesses improve product discovery, increase conversions, reduce returns, and accelerate digital commerce initiatives.
