A year ago, engineers at Snowflake, the American cloud-based data platform, were still spending part of their day on routine tasks, such as scanning dashboards to ensure systems were running smoothly and chasing their colleagues for data to complete trend analyses.
Now, says Qaiser Habib, head of engineering at the Toronto-based company in Canada, AI agents handle much of that legwork, allowing engineers to focus on higher-level decisions.
Recommended stories
list of 4 itemsend of list
Habib spends 20 to 30 hours a week interacting with five AI agents. Snowflake has created agents to review product design or assist on-call engineers during an outage or incident, among other uses. He estimates that the average engineer works with three or four agents daily, using them to carry out coding projects under human supervision.
“There’s no need to bother a human with basic questions anymore,” Habib said, noting that he still collaborates with colleagues on more complex work, such as solving coding problems.
As companies experiment with artificial intelligence agents – systems designed to plan, reason and carry out multi-step tasks – the technology is beginning to reshape office hierarchies in the United States and Canada. Unlike chatbots, which respond to prompts, AI agents can adapt to changing contexts, such as business objectives, and draw on reference tools, including calendars, meeting transcripts, and internal databases, to complete work with limited human supervision.
In some workplaces, AI systems not only complete tasks but also assign them to human workers. As technology improves, AI agents also begin to manage each other. One agent can generate code, for example, while another reviews it for errors and fixes errors before a human approves the final version.
These agent-to-agent workflows can help businesses scale faster. But they also intensify concerns that AI is moving from assisting to supervising and, potentially, replacing jobs.
The most agile office
Anthropic recently expanded access to its cowork agents, allowing users without technical experience to grant Claude, their AI assistant, permission to specific folders on their computers so he can read, edit, create and organize files autonomously.
The increasing use of AI agents is transforming the way organizations operate around the world, even in companies that are not focused on creating technology products. For example, some companies are using AI tools to track performance, recommend promotions, role changes, and even identify roles for elimination.
The change comes as white-collar jobs continue to disappear, particularly in the United States. A host of U.S. employers have announced massive layoffs, primarily affecting entry-level and mid-level workers, and executives have pointed to automation and AI-driven efficiency as part of the justification. When Amazon said in October that it planned to cut about 14,000 jobs, executives cited the potential for AI to help the company operate with fewer layers and greater efficiency. UPS, Target and General Motors also announced steep cuts last year, and this January saw more layoffs than any January in the United States since 2009. Several other companies, including Pinterest and HP, continued to cite AI initiatives as part of the reason.
Goldman Sachs has estimated that 6 to 7 percent of American workers could lose their jobs due to AI adoption, with greatest risks for computer programmers, accountants, auditors, legal and administrative assistants, and customer service representatives. The overall effects on employment, the bank said in August, may be “relatively temporary” as new roles emerge.
The squeezed middle managers
Early predictions suggested that AI would mostly replace entry-level technical jobs, and some experts link recent high unemployment rates among new graduates to AI adoption. But the biggest disruption, said Roger Kirkness, founder of artificial intelligence software firm Convictional in Toronto, is happening at middle management.
Your company’s tools translate executive strategy into operational tasks (a function once handled by supervisors) by delivering daily tasks and feedback to employees through an easy-to-use inbox interface.
In companies of more than 50 people, “where CEOs can’t talk to every manager, our platform continually shows the context that the organization has that is relevant to leadership decision-making,” Kirkness told Al Jazeera.
This does not mean that humans have become irrelevant. But there is growing pressure to reskill, and those who thrive in strategic thinking are better positioned to adapt to AI-integrated work environments, Kirkness said.
“People are basically becoming managers of their previous jobs,” he said, because AI is now capable of performing many of the tasks that were previously within their roles. Instead of completing tasks like coding or designing marketing assets, humans focus on higher-level strategies while monitoring AI systems, he added.
However, recent research indicates that job cuts reflect companies’ anticipation of AI’s potential, rather than its current ability to completely replace human workers.
A Harvard Business Review survey of 1,006 global executives in December found that while AI has played little direct role in replacing workers so far, many companies have already cut jobs or slowed hiring in anticipation of the promised impact.
Most CEOs say they are still waiting for the benefits of AI: 56 percent report no revenue or cost benefits so far, according to consulting firm PwC’s latest global CEO survey of 4,454 executives in 95 countries and territories.
Trust and control
Stefano Puntoni, a behavioral scientist at the Wharton School of the University of Pennsylvania, found that the use of AI is also affecting communication habits in the workplace. Their research shows that employees are often more willing to delegate tasks to AI than to their colleagues, which can help reduce burnout. “There is no social cost,” he said. “You don’t worry about overloading an AI.”
Still, Puntoni maintains that the biggest barrier to adoption is psychological, not technical. Even effective systems can fail if workers don’t trust them. Generative AI, he said, can threaten employees’ sense of competence, autonomy and connectedness.
“If workers feel threatened, they may want the system to fail,” Puntoni said. “At scale, that guarantees failure.”
In other words, implementing AI primarily as a cost-reduction tool may backfire. Layoffs framed as efficiency gains can reduce cooperation and limit the productivity benefits that companies hope to unlock with technology, Puntoni said.
Trust, Kirkness agreed, is the real limitation. To build staff confidence in the tools it sells (and avoid layoffs), Convictional adopted a four-day work week, framing it as a way to share AI-powered productivity gains with employees.
“Mass layoffs in the name of automation destroy trust,” he said.
The human cousin
In the United States, lawsuits have begun for AI-driven corporate decisions, particularly in areas such as denials of insurance claims and alleged AI-enabled hiring discrimination.
Some experts warn that as AI systems become more autonomous, humans risk losing significant oversight and that these agents themselves could become targets of cyberattacks. However, regulation has struggled to keep pace with innovation. Neither the United States nor Canada have clearly defined rules governing AI agents.
Business leaders are testing which functions can be automated and which still require sustained human involvement. For some workers, that uncertainty has become a source of discomfort.
An employee at a multinational company, based in Vancouver, said she sometimes wonders whether the online “coach” used to support employee development is an artificial intelligence system or a human being who relies so much on artificial intelligence tools that the distinction has become blurred. He requested anonymity for fear of professional repercussions.
Some organizations are setting limits. New Ground Wellness, a Canadian clinical counseling and wellness company, uses AI tools like chatbots in its daily operations, but recently rejected a C$20,000 ($14,600) proposal for an AI intake system that would connect therapists with clients.
After receiving feedback from callers, the company concluded that the efficiency gains would not outweigh the potential damage to trust. Their decision also reflects multiple surveys showing a strong preference among Western consumers for human customer service workers.
“We are open to reviewing AI systems in the future,” said Lucinda Bibbs, co-founder of New Ground Wellness, “but at this stage, preserving human connections remains our top priority.”
