.svg)
While the phrases "enterprise AI" and "orchestration" go hand in hand, the execution of these terms has wildly contrasting results. In the rush to implement AI, many enterprises are discovering a gap between the promise of broad role replacement and efficiency gains. The real value enterprises are encountering isn’t in sweeping automation; it’s in the details that enable both humans and AI to perfect the workload handoff between human insight and machine efficiency.
We spoke with Michon Williams, the ex-Chief Technology Officer at Walmart Canada, who has been managing these kinds of disruptions for over two decades. For Williams, today's AI boom isn't a sudden revolution, it’s a wave that's had two decades of input from early digitization to the cloud. This evolution has shaped a new environment for tech leaders, one where assembling disparate solutions is more important than ever. With a track record leading large-scale technology and data initiatives at major retail and financial organizations like TJX, BMO, and RBC, Williams has developed a philosophy for navigating change through the metaphor of a quilt.
In her words, a quilt is the perfect metaphor to describe the practical state of modern enterprise architecture, where the work involves stitching together new tools, legacy systems, and homegrown solutions. It’s an act of curation that accepts imperfection as a given.
A patchwork reality: The metaphor of a quilt captures the human dimension of technology leadership. As Williams explained, just as a quilter carefully selects and arranges fabric to create a design greater than the sum of its parts, "modern CIOs orchestrate teams, AI tools, and processes to produce outcomes that are functional, scalable, and—crucially—aligned with business strategy."
Imperfection is the goal: The patchwork quilt, imperfect yet purposeful, becomes a visual shorthand for the complexity, creativity, and judgment required in leading enterprise AI and platform convergence. "'Quilting' is the reality of modern platforms. You're never going to have the perfect architecture. You're always going to be dealing with scraps of legacy. The CIO's job is deciding which pieces fit into the master quilt and which need to be discarded," Williams said.
The quilt is more than a compilation of data, tech, and process; it’s inclusive of your team too. Williams believes in leveraging communities of practice to help overcome a frequent challenge facing enterprise AI: pilot purgatory. The community model allows an email scanning bot built for lawyers, for example, to be repurposed for finance professionals. Her approach reframes failure not as a setback, but as the price of discovery.
The cost of admission: Williams is clear that early adoption of AI carries both risk and opportunity. "You have to accept that some experiments will fail," she said, "Disruptive technologies like AI demand investment from multiple sources—financial, cultural, and cognitive—and our stakeholders or shareholders naturally expect results."
Loss leaders: Williams believes it’s naïve to bet on every enterprise initiative paying off. In her view, these “loss leaders” aren’t wasted effort; they are experiments that illuminate what works, what scales, and what truly drives value. In Williams' experience, the real advantage comes from staying actively engaged in the experimentation process: observing patterns, learning from missteps, and continuously refining approaches. "Leadership across the enterprise needs to take an active role in evaluating potential successes from dead ends because this process has the potential to reveal breakthroughs that deliver both efficiency and strategic insight," she said.
So how do enterprises take these investments and loss leaders and transform them into something that's capable of generating real ROI? Williams says leaders need to fundamentally change their focus from the tech itself to the evaluation of which tasks belong to machines or humans.
Tasks over titles: Williams believes that effective AI adoption isn’t about handing over entire roles to machines, it’s about deciding which tasks AI should handle and which require human judgment. "We need to stop looking at role automation and really understand tasks and handoffs. That's where the real opportunity for enterprise AI ROI lies," she said.
Bloom's Taxonomy: Drawing on Bloom’s Taxonomy, Williams explained that AI excels at lower-level tasks like knowledge aggregation, freeing human leaders to focus on higher-stakes, strategic decisions. “My biggest concern is when AI is used for synthesis or critical thinking without a human in the loop,” she warned before providing an example of an executive who circulated a ChatGPT-generated strategy only to realize it wasn’t aligned with their intended direction. AI is powerful for accessing and synthesizing information, but humans must provide the critical overlay to ensure outputs are aligned with organizational goals, ethical standards, and bias mitigation. In this approach, enterprises are able to allow technology to amplify human insight rather than replaces it—forging a partnership.
Ultimately, success depends on striking the right balance in the partnership between humans and AI. It means cultivating a culture where technology augments human skill rather than replaces it, where experimentation is encouraged, and where judgment and strategy remain firmly in human hands. "AI is fantastic at aggregating and repeating knowledge, but when it comes to strategy or decisions with human implications, we cannot afford complete cognitive outsourcing. Period," Williams cautioned. At the end of the day, it’s people—not algorithms—who determine which innovations endure, which ideas scale, and which pieces of the enterprise quilt make it into the final product.
