Mise en Place: The Case for Getting Ready Before Going Fast
Holt Renfrew SVP, Information Technology and Supply Chain Alicia Samuel on governance as an accelerant, the luxury brand's carefully calibrated approach to AI, and why "moving fast" means something different when your customers aren't a casual audience.
There's a version of Holt Renfrew's AI story that could easily be misread as caution: two years spent modernizing data infrastructure before customer-facing AI applications were considered, a formal AI squad and cross-functional committee reviewing every initiative, and a deliberate policy of watching others move fast, learning from their mistakes, and only proceeding once the foundations are genuinely ready. But Alicia Samuel, the retailer'sSVP, Information Technology and Supply Chain, would push back on that framing.
"I know that most folks think about governance and they think about going slow," she says. "But that's not the case. It actually accelerates the work that we're doing."
It’s an approach that puts Holt Renfrew in contrast with peers like Groupe Dynamite's David Stevens, who built the company’s data foundation and ran experiments simultaneously. Both approaches are legitimate responses to the same underlying challenge, but what makes Holt Renfrew's path notable is the specificity of the risk calculus behind it, and how clearly that risk calculus flows from the brand itself.
When Your Customers Aren't a Casual Audience
Holt Renfrew is a high-touch, high-trust luxury retailer with a customer base that, as Samuel puts it with some understatement, "is not a casual audience." That context shapes every technology decision she makes.
"One small mistake can really be detrimental," she says. "We focus a lot on our data governance, our security foundation, our access-level controls, and our data loss prevention, governing that part of the organization before we even step into taking a look at what we were going to use from an AI tool and rolling it out."
The customer experience implications are equally specific. Where most retailers would treat an AI-powered chatbot as a straightforward efficiency gain, Holt Renfrew has made a deliberate call not to put one in front of customers. The in-person, relationship-driven experience is the product, and anything that disrupts that feel, however marginally, is a risk that doesn't clear the bar.
"Would we put a virtual bot in front of them? No," Samuel says. "Because we want that personalized touch."
That doesn't mean AI stays out of customer interactions entirely. It means it stays invisible within them— powering recommendations, personalizing communications, and equipping associates with better information before they walk onto the floor. The technology is present, but the customer doesn't see the seams.
Governance as an Accelerant
The framework Samuel built at Holt Renfrew has four components: AI-ready structured data and policy; AI literacy and awareness programs across the organization; architecture and infrastructure capable of handling both deterministic and probabilistic AI capabilities securely; and rigorous business value prioritization that forces every initiative to justify itself against either efficiency, customer experience, or specific KPIs.
An AI squad manages how the organization moves through that framework, and a cross-functional AI committee—which includes Legal and Security from the get-go—reviews policy and use cases as they arise. The structure sounds like it could slow things down. Samuel argues it does the opposite.
Her reference point is striking: "I keep remembering that 88% of AI pilots and POCs don't get to execution," she says. "So that's always been part of my thinking: how do we ensure that we're doing all the right things, learning from everyone else who's gone before us, and establishing the right framework to move us fast?"
It’s like a mise en place in a professional kitchen. Everything chopped, measured, and laid out before the heat goes on, so when those first orders start coming in, the team can move quickly and safely without missing a beat.
Building Literacy Across the Organization
What Samuel has found—and what echoes across the practitioner conversations from MACH X more broadly—is that organizational readiness matters as much as technical readiness. In Holt Renfrew's case, that's meant investing in AI literacy at every level of the business, not just within the technology team.
Six months ago, the organization began distributing licensed access to Copilot across the company. The intent was twofold: get people comfortable with AI tools through direct contact, and use the resulting engagement data to understand where enthusiasm and readiness actually live within the organization. "That helps us define which tools different people need," Samuel says. "And it also drives adoption within the organization."
The signal has been clear and consistent. Samuel says that as the Holt Renfrew executive travels to stores across Canada to share financials, strategy, and awards in person with the broader team, the question that keeps coming from store associates is the same one: tell us about the AI strategy.
"Somebody in the audience will ask, 'What's happening with AI?'" Samuel says. "And then we walk them through the approach and the journey." With the ingredients all laid out, and the kitchen crew ready for the first service, they’ve turned the burners on.
So far, the prep work has paid off. The supply chain team has already demonstrated what that journey can produce, taking hours of manual work out of processes through a handful of agents. It's an early result, but a meaningful one, and it came from within the organization, not from the top down.
The Risk That Keeps Her Honest
For all the structure and deliberateness of Holt Renfrew's approach, Samuel is clear-eyed about where the real risk lives as the organization moves toward more agentic capabilities.
"You don't know what you don't know when it comes to your data," she says. "Especially when we start getting into agents as part of our journey. We are ruthless about making sure our data loss prevention policies are in place. You don't want anything just going out there because one agent is talking to another and there's no human in the loop."
Poor foundations don't just limit what AI can do, they create compounding exposure that becomes harder to manage as systems grow more interconnected and autonomous. For Samuel, the answer isn't to avoid agents, but to arrive at them having done the work that makes them governable.
"This isn't just technical enablement," she says. "It's organizational transformation through embedded technology."
Leigh Bryant
Editorial Director, Composable.com
Leigh Bryant is a seasoned content and brand strategist with over a decade of experience in digital storytelling. Starting in retail before shifting to the technology space, she has spent the past ten years crafting compelling narratives as a writer, editor, and strategist.