At Shopify, AI is scrambling the org chart. Interns aren’t only there to learn from experienced engineers. They’re there, in part, so senior leaders can learn from them.
The company recently expanded its internship program from about 100 students to more than 1,000. Farhan Thawar, Shopify’s VP and Head of Engineering, says AI is part of the reason: students now arrive having used AI and treat it as a normal part of how they work. Rather than concentrating interns in a specialized unit, Shopify distributes them across teams so their habits, questions, and ideas can travel. When company leaders and interns are in the office together, Thawar says, senior people “pick their brains.”
The point isn’t that interns suddenly know more than senior engineers. It’s that no level of the organization holds all the expertise AI adoption requires. Fluency with emerging tools may sit with the newest employee, professional judgment with the most experienced, and the best application with the person closest to the work. To benefit, though, organizations must find knowledge wherever it emerges, combine different forms of expertise, and turn what team members know into something others can use.
Shopify isn’t an outlier. Across our conversations with Canadian employers and people managers, we’ve found the same reversal: new hires are increasingly expected to act as internal change agents, helping more senior staff understand and adopt AI. The pattern is visible nationally and internationally, too. The International Workplace Group found that 55 per cent of younger workers are actively helping their older colleagues learn and use AI tools.
This reframes the challenge facing employers. The question is no longer how to train everyone. It’s how to build organizational pathways for knowledge to travel, opportunities for judgment to develop, and systems that allow the lessons from experimentation to spread. That’s the difference between an organization with AI-capable employees and an AI-capable organization.
1. Build pathways for knowledge to travel.
TELUS has built a pathway. Rather than relying on a central technology team to anticipate every useful AI application, the company gives employees access to Fuel iX, a secure platform where technical and non-technical team members can create AI copilots for particular tasks and workflows. By November 2025, 70,000 employees had access to the platform and had created more than 21,000 custom copilots. TELUS CIO Hesham Fahmy described the result as a “groundswell of ideas” across the organization.
What’s significant isn’t that TELUS employees are building lots of tools. It’s that someone close to the work can identify an opportunity and turn that insight into something another team can use. Useful knowledge doesn’t have to originate with a senior leader or central AI team, and it doesn’t have to remain with the person who discovered it.
That’s how individual AI fluency starts becoming organizational capability. But making a useful practice portable is not the same as developing judgment. A copilot can carry a prompt, process, or workflow to other employees. It can’t, by itself, teach them how to recognize a good result, weigh competing considerations, or know when AI shouldn’t be used at all. That kind of expertise develops differently.
2. Build judgment through real work.
If AI can make a junior employee productive faster, it can also make it harder for that employee to become an expert. Early-career work has always doubled as training. Tasks that look routine to an experienced professional are often where a new worker learns to recognize quality, catch mistakes, make trade-offs, and recover when an approach fails. When AI moves directly from question to answer, it can bypass the trial, feedback, and correction loop through which professional judgment develops. The immediate work may get faster even as the human capability beneath it gets weaker.
Deloitte Canada identified this “capability erosion” as a system problem. Its response was to make apprenticeship a more deliberate part of how work gets done. Experienced practitioners work alongside emerging talent on live assignments, making their expectations, methods, and judgment calls visible. Junior employees don’t simply receive a completed answer; they observe how experienced colleagues frame a problem, weigh competing considerations, respond to mistakes, and determine what good work looks like.
The lesson isn’t that organizations should preserve routine work for its own sake. It’s that when AI removes a task, leaders must ask what employees once learned from doing it and how they will develop that capability another way. Courses can teach employees what a tool does. But judgment develops through real decisions, visible reasoning, feedback, and correction. Without that redesign, organizations may gain productivity today while weakening the expertise they will need tomorrow.
3. Turn experimentation into organizational learning.
Practice develops individual judgment. But experimentation alone doesn’t make an organization smarter. If every employee’s experiment remains isolated, colleagues must rediscover the same uses and repeat the same mistakes. Organizational capability grows when people can test ideas close to the work and what they learn shapes what happens next. RBC Capital Markets has created that opportunity through Aiden, its enterprise AI platform.
Aiden is available to all 8,000 Capital Markets employees, with tools that let them create agents for specific tasks. Employees across the business have already built thousands. Experimentation is distributed across the organization rather than confined to a central AI team. RBC pairs that bottom-up activity with a business-led process for deciding what to advance. A cross-functional team develops solutions to high-value business problems on a common, governed platform.
After demonstrating value, AidenBanker moved from pilot to production and is now being rolled out across Global Investment Banking. This is how experimentation becomes organizational learning: employees create new applications, evidence determines what scales, and successful ideas become part of the institution’s infrastructure.
From AI fluency to AI capability
When the newest hire becomes the AI teacher, the larger lesson isn’t about age or seniority. It’s that expertise no longer moves predictably along the arrows of the org chart. The real test of AI readiness isn’t how many employees complete training, but whether knowledge moves, judgment develops, and successful experiments spread.
Canada’s AI readiness will depend on more than producing AI-capable workers. It’ll depend on building organizations capable of learning from them.
BHER is advancing this work through its AI Leadership Table. If you’re a company or post-secondary institution that wants to help us address talent, training, and coordination challenges in Canada, please reach out to Valerie.Walker@bher.ca about BHER membership.
Download This Thought Leadership Piece
An Action Plan for Canada's Next Economy
Canada has entered a critical period that will shape its economic strength, technological leadership, and national sovereignty. At BHER’s 2026 Executive Summit, leaders identified six priority areas where coordinated action across governments, industry, and post-secondary can unlock Canada’s next phase of economic growth and strategic capacity.
