As executive ‘cognitive twins’ emerge, key leadership roles must remain human

Sam Rosenbaum, Christoph Schulze and Mike James Ross are consultants and experts at Egon Zehnder in Toronto and Montreal.

Earlier this year, a LinkedIn post from a software engineering executive we know stopped us mid-scroll. The executive was announcing the launch of his “cognitive twin”: a “multi-agent AI system that thinks like me, advises like me and can handle certain questions when I’m not in the room … Still rough, still evolving … but working.”

The code for this project has since been open-sourced, meaning anyone can use it to build their own twin. Sure, tech moguls had been experimenting with AI twins for a couple of years, but this felt decidedly closer to home.

As professionals in the business of executive talent and leadership, the prospect is both fascinating and horrifying. The leverage this technology might offer time-starved executives is obvious. Equally clear are visions of a dystopian workplace in which employees interact more often with an executive’s artificial intelligence clone than with the actual executive.

Cognitive twins have moved beyond the realm of science fiction. At their simplest, they are large language models trained on an individual’s writings, decisions, communications, expertise and experiences. Advances in multimodal AI mean they can manifest as true executive avatars, learning an individual’s appearance, body language, facial expressions and voice. Their purpose is not merely to store what a person knows, but to approximate how that person thinks and communicates.

AI-powered executive avatars are already taking seats around the digital boardroom table. The CEOs of tech companies Zoom and Klarna have used AI avatars to deliver portions of prepared earnings calls. Others, including LinkedIn founder Reid Hoffman, Otter.ai CEO Sam Liang and Accenture managing director Iliana Oris Valiente, are entrusting their avatars to handle unscripted communication, bringing the technology closer to a fully realized “cognitive twin.”

These systems offer compelling benefits for organizations. Time-strapped executives are often a bottleneck to decision-making. Without access to leaders, decisions can be slow, inconsistent or flawed.

Cognitive twins could help unblock workers by providing on-demand guidance and even making lower-stakes decisions without an executive’s direct involvement.

Bobby Morrison, former Chief Revenue Officer of Shopify and advisor to several AI-native startups, has experimented extensively with executive cognitive twins.

“I think of this as democratized ‘founder mode’,” he explained when we spoke recently. “The real opportunity isn’t simply creating a digital twin of an executive; it’s encoding the leadership team’s context, company values, first principles and decision frameworks so people throughout the organization can make better decisions without waiting for the executive to be in the room. The best cognitive twin eventually makes the executive less of a bottleneck, not more omnipresent.”

Another opportunity is executive access and employee engagement. Many of the executives we work with express a desire to spend more time interacting with their teams across time zones: communicating vision and strategy, answering questions and gathering feedback. Deploying cognitive twins to handle some of this communication could increase executive visibility and drive greater employee engagement. One-to-many, town hall-style communication could be replaced by thousands of virtual coffee chats that deliver consistent messaging at a fraction of the effort.

Finally, cognitive twins may help preserve the institutional memory of aging executives. The overall number of retirements in Canada, measured as a 12-month average, increased to 276,800 in August 2025 from 183,900 in August 2012, a 50-per-cent gain. Cognitive twins could preserve both an executive’s knowledge and the personal context that often brings it to life.

The opportunity is real, but so is the risk. Cognitive twins may become powerful management tools by sharing information, accelerating decisions and preserving expertise, but they are not substitutes for leadership. The better AI becomes at reproducing past thinking, the more valuable human leadership becomes.

This paradox is best understood through two distinctions. The first is between knowledge and judgment. Knowledge can be captured, rearranged and reproduced. AI systems are excellent at this.

Judgment is different. It requires leaders to navigate novel and ambiguous situations, weigh competing interests and make choices informed by values and ethics. AI can be a useful thought partner, but fully delegating judgment to an executive cognitive twin risks creating a “nostalgia machine” that unthinkingly institutionalizes yesterday’s assumptions.

When a situation requires a simple transfer of knowledge, an AI delegate may be appropriate. When a decision involves ambiguity, values or meaningful consequences, there is no substitute for human judgment.

“It’s a point of view that I’m taking into my decisions, and then I’ll ultimately use judgment to determine what the right course of action is,” Morrison explained.

The second distinction is between management and leadership, popularized by Harvard Business School professor John Kotter in his 1990 article “What Leaders Really Do.” Kotter argues that management is the work of planning, organizing and problem-solving while leadership is the work of aligning, motivating and inspiring.

Distributing information widely and consistently takes skill and time, and AI can help. But while information is easily distributed at scale, trust and followership are not. Leaders can use cognitive twins to extend their reach, not outsource human intimacy.

“The executive skills that get elevated are effective communication, meaningful, genuine relationships and empathy,” Morrison reflected.

Could a CEO’s cognitive twin deliver a routine company town hall? Probably. But it is difficult to imagine a corporate battle cry, a tough transformation announcement or heartfelt words of appreciation meaning much coming from anyone other than a human leader. Cognitive twins may help executives manage more efficiently. Building commitment and bringing people along, particularly in times of uncertainty, remain acts of human leadership.

AI-powered cognitive twins are likely to become more widespread. Organizations should embrace their capacity to increase engagement, accelerate routine decision-making and preserve expertise. But leaders must draw clear boundaries: use AI to share knowledge and support decisions but retain human responsibility for judgment, trust and change.

The leaders who thrive in the age of cognitive twins will not be those who create the most convincing AI versions of themselves. They will be those who know which parts of them can be replicated and which parts should never be.

This column is part of Globe Careers’ Leadership Lab series, where executives and experts share their views and advice about the world of work. Find all Leadership Lab stories at tgam.ca/leadershiplab.