Enterprises are investing heavily in AI, but many face challenges due to automating outdated processes and a lack of strategic clarity.
Liz Miller of Constellation Research joined us for an interview to discuss the need for a CEO-led digital transformation and the role of a Chief AI Officer.
Building a knowledge layer that integrates structured and unstructured data is crucial for enterprise-wide intelligence.
Success in AI requires a commitment to becoming a digital organization, not just implementing AI technologies.
Enterprises are spending billions to explore intelligent automation, but the path to true ROI is often littered with expensive pilot projects, frustrated teams, and a growing sense of strategic confusion.
We spoke with Liz Miller, VP & Principal Analyst at Constellation Research, Inc., where she advises leaders on navigating the intersection of marketing, CX, and business strategy. For Miller, the reason so many AI initiatives are merely “lighting money on fire” is that leaders are automating broken processes, driven by a profound misunderstanding of what AI is and a culture that punishes the very questions it’s meant to answer.
Faster old tech: “If you simply automate an old process where you already know the outcome, applying a new AI technology to it will only result in faster old technology.” The shortfall, according to Miller, has little to do with the underlying tech itself, but with a fundamental disconnect starting in the C-Suite.
Miller argued that in many organizations, “data is a cause for punishment”, meaning hard metrics can be used against the employee if they are deemed insufficient. This creates a climate of fear where workers are terrified to ask the deep questions that could uncover uncomfortable truths and also unlock massive value.
The ‘don’t touch my button’ game: This fear cascades into political dysfunction, a corporate game of data hoarding that sabotages any hope of cross-functional intelligence. “No one wants to lose the game of corporate ‘Don’t touch my button.'” According to Miller, the CISO, CIO, and CTO are often locked into conflicts of indecision, what she called “The Land of ‘No'” that make it difficult to broker the authority needed for true AI orchestration.
According to Miller, the solution begins with a CEO-led decision to become a digital-first company, and “not just a company that does digital things”. Miller stressed the commitment must be embodied in a new, empowered role of a Chief AI Officer.
Enter the CAIO: “The chief AI officer is only going to be as successful as the CEO’s mandate to be a digital organization,” Miller said. “This isn’t another technology role. It is a business operations role that looks at where the business transforms and how people, process, and platform must transform together.”
A new cultural contract: The role cannot be allowed to devolve into a “weird bully stick for operational optimization,” Miller warned, and it must confront the financial reality that AI is not an automatic cost-saver. “Replacing one employee with 90 more digital agents can easily backfire without strategic oversight,” Miller said, clarifying that the best CEOs form a new cultural contract where every functional C-Suite leader has the same room to fail and succeed.
With the right leadership in place, the focus shifts to building a new kind of data infrastructure. But Miller offered a warning against mindless accumulation.
The data decathlon: “We are not running in a data decathlon, and the person handling the greatest mountain of accumulated knowledge will not be the medal winner” she said. The true prize is an enterprise-wide “knowledge layer” built on a foundation that can fuse structured and unstructured data.
Creating the knowledge graph: “What we’re actually in search of in this new age of AI is a knowledge layer that feeds the entire enterprise. That data should feed into a knowledge graph that feeds the entire organization and it’s fed from the entire organization, so it’s constantly refreshing.” This creates a powerful flywheel, allowing the business to move from asking “what happened?” to understanding “why?” and finally, to proactively predicting the next “why not?”.
Ultimately, success demands more than just deploying new models and agentic frameworks. It requires the discipline to choose the right tool for the job. “You don’t have to decide to be an AI organization. You can just do ‘AI things,'” Miller concluded. “But you can’t decide to do ‘AI things’ unless you’ve first decided to be a digital organization. You just don’t get to morph the two.”
