A year ago, engineers at Snowflake were still spending chunks of their day on the unglamorous stuff — scanning dashboards, chasing colleagues for data, waiting on answers to routine questions. Today, much of that groundwork is handled by AI agents, freeing people to focus on higher-level decisions.
Qaiser Habib, Snowflake's Toronto-based head of Canada engineering, now spends 20 to 30 hours a week interacting with five different AI agents. His team uses them to review product designs, support engineers during outages, and supervise coding projects. The average engineer, he says, works with three or four agents every single day.
"You don't have to bother a human for basic questions anymore," Habib said.
That quiet shift is playing out in offices across North America — and the implications go far beyond saving time on admin work.
From chatbots to autonomous agents
Most people are familiar with AI chatbots: you ask a question, you get an answer. AI agents are different. They don't just respond — they plan, reason, and execute multi-step tasks with limited human oversight. They can draw on calendars, meeting transcripts, internal databases, and business goals to get things done autonomously.
More significantly, agents are beginning to work together. One might generate code while another reviews it for bugs, with a human only stepping in to approve the final version. These agent-to-agent workflows let companies scale faster than ever — but they also raise a harder question: at what point does AI move from *assisting* workers to *supervising* them?
The squeeze on middle management
Early predictions pointed to entry-level technical jobs as the first casualties of AI adoption. But Roger Kirkness, founder of AI software firm Convictional, sees the bigger disruption happening one layer up. His company's tools translate executive strategy directly into daily operational tasks — a function that used to belong to middle managers.
"People are basically becoming managers of their prior jobs," Kirkness said. Instead of writing code or designing marketing assets, workers are increasingly focused on monitoring AI systems and making higher-level strategic calls.
This doesn't mean humans are becoming irrelevant. But the pressure to reskill is real, and it's growing.
Job cuts today, or anticipation of tomorrow?
The numbers are hard to ignore. Amazon announced plans to cut roughly 14,000 jobs, citing AI's potential to help the company operate with fewer layers. UPS, Target, and General Motors announced deep cuts. January of this year saw more layoffs than any January since 2009. Goldman Sachs estimates 6 to 7 percent of U.S. workers could eventually lose their jobs to AI — with particular risk for programmers, accountants, legal assistants, and customer service reps.
But here's the nuance: a Harvard Business Review survey of over 1,000 global executives found that AI has played little direct role in replacing workers *so far*. Many companies have already slowed hiring or cut headcount — not because AI has proven its worth, but because they expect it to. A separate PwC survey found that 56 percent of CEOs report no revenue or cost benefits from AI yet.
In other words, a lot of the disruption happening right now is driven by anticipation, not results.
The trust problem nobody's talking about
Stefano Puntoni, a behavioral scientist at Wharton, has found something counterintuitive: employees are often *more* willing to delegate to AI than to human colleagues. "There's no social cost," he said. "You don't worry about burdening an AI."
But Puntoni also argues that the biggest barrier to AI adoption isn't technical — it's psychological. When AI threatens workers' sense of competence and autonomy, they may quietly root for it to fail. At scale, that resistance can undermine the very productivity gains companies are hoping to unlock.
The lesson? Deploying AI primarily as a cost-cutting tool can backfire. Companies that use layoffs to signal AI efficiency may find they've traded long-term productivity for short-term savings.
Kirkness has taken a different approach. To build trust in the tools his company sells — and to avoid layoffs — Convictional adopted a four-day workweek, framing it as a way to share AI-driven gains with employees.
"Mass layoffs in the name of automation destroy trust," he said.
Where humans still win
Not every company is sprinting toward full automation. New Ground Wellness, a Canadian counselling and wellness firm, recently turned down a $14,600 proposal for an AI-powered intake system that would match therapists with clients. After hearing from callers, they concluded the efficiency gains weren't worth the cost to human connection.
"Preserving human connections remains our highest priority," said cofounder Lucinda Bibbs.
Their instinct is backed by data: surveys consistently show that Western consumers strongly prefer human customer service workers, especially when the stakes feel personal.
AI agents are no longer a future-tense conversation. They're already in the office, taking on tasks, collaborating with each other, and in some cases, assigning work to humans. The question isn't whether this technology will reshape how we work — it already is. The question is whether companies will use it to elevate their people or simply reduce their headcount.
The organizations getting it right seem to understand something important: the technology only works if the humans around it trust it. And trust, it turns out, isn't a feature you can automate.