Earlier this year, an employee left a job at Klue Labs Inc. The role involved writing help documents for customers of the Vancouver company’s business intelligence software, which provides insights to corporate clients about competitors. Klue was about to post the vacant position when something occurred to chief executive officer Jason Smith: He didn’t have to hire anyone. The company could try to use generative artificial intelligence, specifically an agent that can handle a series of tasks, to automate much of the work. Maybe not with the same results as a real person, but close enough.
That prompted him to do something drastic. He knew how adept AI tools were becoming for research, writing, coding and other tasks, which both excited and unnerved him. The technology could completely upend his company. These days, a couple of dropouts can grow a business fast with nothing more than a good idea and a subscription to some AI coding tools that can whip up prototypes and basic applications. “If I didn’t make a dramatic move, at some point I wouldn’t have the choice,” Mr. Smith said later.
One morning in June, at the company’s all-hands Zoom meeting, he told employees that in about a week, up to half of them could be laid off. The company would offer voluntary buyout packages, while some employees would be cut. Those who chose to stay would be expected to embrace AI and do more with it. Klue wasn’t running out of cash, but it had to reinvent itself to stay competitive and inject AI into every aspect of its operations, he believed. Klue let go of 85 people, about 40 per cent of the company, with most employees leaving voluntarily. No department was untouched.
As part of a long term AI adaption strategy, Jason Smith, CEO and co-founder of Klue, announced mass layoffs in June.Kayla Isomura/The Globe and Mail
Since then, more than a dozen CEOs have been in touch with Mr. Smith to applaud the bold move and confess that they, too, have been considering something similar. They mostly wanted to know his calculus for how deep to cut.
While not every CEO is about to shed huge numbers of employees, they are definitely all thinking about AI. It’s often pointed out that adoption is low among Canadian companies, and that’s true. It’s also pointed out that generative AI is too error-prone to be useful. That’s not true. Companies are finding they can get lots of mileage out of tools that may never be 100-per-cent reliable.
Among more than a dozen CEOs and executives interviewed by The Globe and Mail, there has been a notable shift this year in approaching AI. The technology has long been hailed by executives, even before the launch of ChatGPT in 2022. But now, instead of tentative pilot projects and nudges to employees to try it out, some companies are bringing AI tools to everyone, and mandating that employees use them – or else.
The newfound urgency, stemming from the promise of productivity gains, competitive pressures and the fact that some tools are simply better than they were a couple of years ago, is prompting some executives to think deeply – and talk openly – about how many employees they’re going to need in the years ahead. Layoffs are difficult and come with a human cost, but can now be seen in a different light. “A layoff actually is a good thing in an AI-first world,” Mr. Smith said. A healthy company, he said, is no longer one that is merely adding people, but one that’s growing with fewer people.
Tech firms such as AlayaCare are growing revenue without adding employees; Hopper Inc. and League Inc., meanwhile, are doing the same while head count has shrunk. Startups are using AI coding tools to develop software faster and with fewer employees, allowing them to turn a profit sooner than ever before, and without as much need for outside financing. This breed of startup poses a threat to legacy companies, helping to explain why some CEOs are nervously looking over their shoulders while strong-arming employees to get more work done with AI.
AlayaCare, headed by CEO Adrian Schauer, is among the tech companies growing revenue without adding employees by incorporating AI into its operations.Andrej Ivanov/The Globe and Mail
It’s not just tech companies that are urgently incorporating AI. Established corporations such as Manulife Financial Corp. and Royal Bank of Canada are rolling out AI tools to wider swaths of employees. At the bank’s investor day in March, executives gushed about all the things the company can do with generative AI; nearly two years ago, CEO Dave McKay said it wasn’t “ready for prime time.”
Generative AI is manna from heaven for executives. More productivity, lower costs and fewer employees? That’s how you make a CEO weak in the knees. For just about everybody else, AI is a potentially destabilizing force. “This is a hugely disruptive technology,” Nick Frosst, co-founder of Toronto-based AI company Cohere, said at an event in June. “I am worried that it will exacerbate the trends that we are seeing already in the economy, of increasing income inequality and bad jobs.”
That outcome is not inevitable. A lot depends on a choice facing CEOs, boards and investors. They can harness the benefits of generative AI to grow faster, chase more business and, yes, hire or free up people to support new opportunities. They can also slash jobs and squeeze more work out of skeleton crews.
The cynic in you might presume to know the answer. But in truth, we don’t know which way they’ll go.
This spring, Michael Serbinis, founder and CEO of Toronto-based tech company League, made it clear to his employees that AI use was non-negotiable.Fred Lum/The Globe and Mail
Michael Serbinis had a gnawing feeling: His company wasn’t moving fast enough to integrate AI. In April, he typed out a memo to his Toronto-based tech company, League. The e-mail included a note that it was “written by Mike, edited by AI” and made clear that AI use for employees was non-negotiable. “If you are a leader here at League, you need to have a conversation with your team ASAP about how to get to 100%,” he wrote.
League, which develops online health care platforms used by benefits providers and others, is reworking job descriptions to account for AI and will evaluate employees’ AI usage in performance reviews. Mr. Serbinis, meanwhile, is thinking about head count. The company already employs fewer people than it did two years ago, while revenue has grown. “There’s just certain roles we will probably not ever hire for again,” he said. That includes some customer service jobs and those related to content creation.
Geotab Inc., which employs 2,400 people and makes technology for companies to track vehicle fleets, is pushing adoption harder, too. “We started ratcheting up the temperature,” said Neil Cawse, the Oakville, Ont., company’s CEO. Geotab uses Google’s chatbot for day-to-day tasks such as writing customer proposals, while coding assistants are widespread. Geotab has AI tools for its hiring process, while AI has reduced the time to review some legal documents. These days, the company is exploring AI agents to handle video creation and repetitive tasks in its human resources department.
Employees have asked Mr. Cawse whether they’re about to be replaced by AI. He has assured them that’s not the case. Instead, he’s told them they will be replaced by people who use AI. The sentiment has become a truism of late, but he really means it: Geotab will assign an AI score to employees. “We will replace people who are not using AI,” he said. “We have no choice.”
Neil Cawse, the founder and CEO of Geotab, is among the business leaders pushing AI adoption at his company, from day-to-day tasks to its hiring process and beyond.Glenn Lowson/The Globe and Mail
One result of these changes is that some companies are more reluctant to hire. The CEOs of Shopify Inc. and Open Text Corp. said earlier this year that they will look to use AI to complete work first before bringing in new employees. More companies are embracing that philosophy. “If you’re going to ask for head count, you better really have demonstrated that your team is leveraging all the tools,” said Adrian Schauer, CEO of AlayaCare in Montreal, which sells software to home-care providers.
AlayaCare’s work force has remained at around 560 people the past three years, while revenue has doubled. The only department still asking for more people is software implementation, which involves helping customers get AlayaCare’s platform up and running (though Mr. Schauer is trying to take the “human effort” out of that work, too.)
When it comes to developing the software itself, he doesn’t need additional bodies, especially entry-level coders. “The appetite for bringing on juniors continues to go down,” he said. “A lot of the work you would give to a junior, you can give to your AI copilot.”
Canadian tech companies have gone through a reckoning in recent years, as rising interest rates have put more focus on turning a profit, prompting layoffs in some cases. Montreal-based online travel company Hopper laid off 30 per cent of its staff in fall 2023, but AI is helping it stay lean.
“Our revenue is growing,” said Hopper CEO Fred Lalonde, while “all of our human numbers are going down.” When employees leave, Hopper doesn’t have to fill those roles as much as in the past. Hopper now employs less than 600 people while revenue has grown about 30 per cent each year. “It’s a decision to say, we want to be a company that has the most amount of automation,” Mr. Lalonde said.
The marketing team at Clutch Technologies Inc., a Toronto-based online used car retailer, relies on generative AI to churn out initial versions of ad images and copy. “We had content folks in-house. We have less of them now,” said CEO Dan Park, who added the remaining team can work a lot faster. “We used to spend thousands of dollars a month on freelance writers, but now we spend much less.” Clutch still relies on real photoshoots with real human beings – for now. “We’ll see where that ends up,” he said.
Clutch founder and COO Steve Seibel, left, and CEO Dan Park. The Toronto-based online used car retailer relies on generative AI to churn out initial versions of ad images and copy, but still uses real photoshoots.JENNIFER ROBERTS/The Globe and Mail
Anwaar Malik’s company could not exist without AI. A former private wealth manager, he built a tool last year that uses the technology to analyze financial data to inform his personal trading strategies. While his creation proved popular with retail investors, he instead focused his Waterloo, Ont.-based company, AllMind AI, on hedge funds. He launched the product in May, and a handful of customers use it to write initial drafts of in-depth research reports that would ordinarily take days.
More than 80 per cent of the $400,000 AllMind raised last year is still in the bank, because Mr. Malik doesn’t need many employees. The company employs four people, including Mr. Malik, and he’s looking to hire someone with experience focused on machine learning. “If you give a junior engineer an AI tool, they’re still probably going to produce junior level code,” he said. A more senior employee will not only produce better work, but will do so faster with AI.
These trends are creating a nearly unprecedented situation with tech startups: They’re able to get products to market and acquire customers faster and with fewer employees. Some companies can turn a profit after raising a single round of startup capital. “Ten years ago, it was almost unheard of for a tech company to just raise a seed round,” said Janet Bannister, founder of venture capital firm Staircase Ventures. Because of AI, she said, “they may not even need to raise again.”
At the other end of the spectrum, Royal Bank started rolling out an AI software platform to select employees in July called North, the flagship enterprise offering from Cohere. There are a few ways employees will use North at first. The software can pull information from multiple data sources to compile reports for wealth managers and investment advisers to prepare for client meetings. “It takes hours to do something like that with humans,” said Foteini Agrafioti, RBC’s chief science officer.
Nick Frosst, co-founder of Toronto-based AI company Cohere. North, the company’s flagship AI software platform, has recently been rolled out to RBC employees.Christopher Katsarov /The Globe and Mail
North can also write research reports on companies covered by equities analysts that will then be reviewed by a human before sending them to clients. “You can only cover so many stocks because you’re limited by time,” she said. “This helps us accelerate that process and cover more stocks with the same work force.”
Cohere had limited North to a small group of customers but made the product widely available this month. “I see North as the union of everything we’ve been building,” said Cohere co-founder Nick Frosst. A lot of companies have struggled to move beyond AI pilot projects, he said, whereas North is powered by Cohere’s large language models and built to scale across an organization – quickly. The platform can access a company’s data and tap into other tools to answer questions, prepare documents, summaries, tables and charts – the lifeblood of enterprise, for better or worse – and can be used to create custom agents to handle repetitive tasks. “By augmenting and automating the bureaucratic and the boring, we can enable people to do the things that they’re particularly good at,” he said.
At Manulife, many employees use a custom chatbot introduced in 2024 called ChatMFC to translate legalese into plain English, work on performance reviews, and write the first drafts of marketing copy and other documents. This year, Manulife is using AI to speed up the underwriting process – first to generate a quick insurance quote, and then assess documents, extract important information and summarize it for the underwriters.
Manulife is also developing tools for portfolio managers to pull in financial data, news and government announcements so they can ask detailed questions about the impact on companies and stocks. “Nobody could do this before,” said global chief AI officer Jodie Wallis. “This really allows us to make much better decisions, much more quickly.”
Klarna’s headquarters in Stockholm, Sweden. The consumer finance company tried to replace its customer service reps with AI, but found many of its customers preferred speaking to humans rather than chatbots.Supantha Mukherjee/Reuters
What could be lost in this relentless productivity drive? Let’s start with jobs – the ones people have today and positions that might not materialize in the future. About 60 per cent of workers are “highly exposed” to generative AI, according to a 2024 Statistics Canada study. Unlike other waves of automation, the affected positions are predominantly white-collar jobs that require higher education. These people are not about to be kicked to the curb, though. “Employers may not immediately replace human labour with AI, even if it is technologically feasible to do so, because of financial, legal and institutional constraints,” the authors wrote, adding many roles will benefit from the technology.
Some companies that attempted to swap out employees for AI have backtracked. Swedish consumer finance company Klarna tried to replace customer service reps with AI, but it turns out we still prefer talking to humans, not chatbots.
If companies stop or severely curtail hiring, however, that could be felt by job seekers. “There’s going to be a whole bunch of people in the ecosystem that can’t find work,” said Alok Ajmera, the CEO of Mississauga-based Prophix Software Inc., which sells digital financial planning and analysis tools. He’s not advocating for that approach, however, and it could prove short-sighted. A company fixated on using AI solely to cut costs or never hire again leaves itself vulnerable to a competitor that is focused on using AI to expand faster. “This technology is available to everyone,” he continued. “To stay competitive, I still have to continue to scale.”
Looking to generative AI as a way to grind down the work force represents a failure of imagination. “Not gonna happen on my watch,” said Don Murray, CEO and co-founder of Safe Software Inc. in Surrey, B.C., which provides spatial data for governments and other clients. “We have so much we can do.” He is pushing adoption hard, recently telling employees that everyone, no matter the role, is expected to use AI daily. He’s a heavy user, too; if he wakes up at 2 a.m. with an idea, he’ll tell it to ChatGPT.
Don Murray, CEO and cofounder of Safe Software Inc. in Surrey, B.C., says he is pushing AI adoption hard and expects every employee to use AI on a daily basis.Jennifer Gauthier/The Globe and Mail
Even so, he acknowledged that hiring could slow down in some parts of the company, such as customer success. These employees work directly with clients to field questions and deal with issues. AI might be able to handle some of these queries, Mr. Murray said, freeing up employees to work on harder problems. Or they can write help documents based on new questions from clients. “You’re creating great new knowledge content that customers can find, because AI can’t create knowledge,” Mr. Murray said.
Moreover, where AI can relieve a bottleneck in one area, it may create another one elsewhere. Anwaar Malik of AllMind AI may not need an army of coders to write software right now, but he does need people to market that software. He doesn’t want to rely on AI for advertising. “No matter how well AI can write a piece of code, it cannot generate text that is worth reading,” he said.
While a lot of fear is focused on how AI may replace jobs, there are subtler but equally concerning effects. Matt Beane, an associate professor at the University of California, Santa Barbara, published a book last year called The Skill Code that explores how technology can break the relationship between experts and novices, threatening how skills are transferred. The book was researched before generative AI exploded, and some of his work focuses on how surgical robots can stunt the development of young surgeons.
A team of surgeons, nurses and surgical fellows at Toronto General Hospital prepare for a simulation training session using Medtronic’s Hugo Robotic Assisted Surgery (Hugo RAS) system.Supplied
Generative AI threatens to do the same, to more workers, all at once. “If you make novices optional in the work, you’re breaking the millennia-old treadmill that produces that next wave,” Prof. Beane said.
Since he published the book, he’s heard from people across industries, all worried about this phenomenon. How does the junior lawyer learn if the grunt work typically thrown at them – document review, preparing briefs and so on – can be handled by a senior lawyer with AI? What about the young analyst who finds that spreadsheets and financial models can be populated by an AI agent that will also assemble a fancy slide deck? Does the entry level programmer truly understand the code that AI generates? Young people can’t jump to the higher-value intellectual work performed by senior employees.
“They aren’t experts in the work itself,” Prof. Beane said. “That kind of wisdom comes from digesting the complexity of your work.” It’s a problem companies should attack with as much urgency as the zealous drive to use AI. “There’s a brilliant opportunity right now, and it’s to find those ways of organizing that are jointly optimized for skill development and productivity,” he said.
These issues are discussed at Dentons Canada LLP, as the law firm looks to deploy tools more widely. Many lawyers use Harvey multiple times a day, a specialized AI program for the legal industry that can analyze contracts, and assist with due diligence and regulatory compliance, among other chores. “If the simple tasks are done by AI, that breaks our training model,” said Andrea Johnson, the firm’s global co-head of private equity. “If you press a button, which auto-generates a document, what are your skills?”
The work of junior lawyers hasn’t been affected by AI yet, she said, though the firm is watching this closely. In fact, the AI tools Dentons has today are helpful, but are not delivering the “step change” that she believes the technology one day could.
Trade-offs may be discovered in time. Ms. Johnson wondered about due diligence during mergers and acquisitions, in which one company pores over the intimate financial and operational details of another before completing the transaction. AI can speed it up by analyzing documents and spotting regulatory and legal issues. “Is there actually a danger in your M&A process becoming too fast?” she asked. Determining a culture fit during that whole ordeal is important, too. “If it’s too smooth and you don’t have the usual impediments, you won’t necessarily have seen people deal with conflicts,” she said.
Ms. Johnson’s colleague Nicole Miles and her team use generative AI to help research and produce marketing content for the Dentons website and other platforms, which has changed how they work. “We used to sit around and brainstorm, and I do kind of miss that,” said Ms. Miles, chief clients and markets officer for Canada. Now, they plug a prompt into an AI application and start from there. “But we’re able to produce so much more, so much faster,” she said.
Since Mr. Smith made the drastic move to reduce his workforce and turn towards AI tools earlier this year, more than a dozen CEOs have been in touch to applaud the bold decision as they too consider taking similar steps.Kayla Isomura/The Globe and Mail
Klue, with its reduced work force, could be among the first Canadian companies to learn the benefits and pitfalls of going all-in on AI. So far, Klue’s Jason Smith has no regrets, and anticipates his company will hire again, when needed. His decision is not a bet that AI will suddenly become superintelligent; it’s a response to what the technology can already do. If it improves, all the better.
One development he’s watching for is advanced AI reasoning capabilities, so that applications can analyze information and draw insightful conclusions, just like we do. “Until AI has reasoning at that human creativity level, we’re going to need humans to drive a new way of thinking,” he said.
Just maybe not as many as before.