Making AI work for your business is all about picking the right tasks

Globe and Mail

5 hours ago

Ellen Hyslop knew for artificial intelligence to be useful it had to boost efficiency, without dulling creativity.
At Canadian sports media company The Gist, where she and her team are rethinking day-to-day operations, AI is being adopted with a clear purpose: 'to do things that we need to be doing in the most efficient and useful way possible [without compromising] the quality and creativity of the work that we do,' she says.
Founded in 2017 by Ms. Hyslop, Roslyn McLarty and Jacie deHoop, The Gist is a women-led company that offers equal coverage of men's and women's sports via newsletters and a podcast. Last year, the brand hit one million newsletter subscribers, a milestone that Ms. Hyslop says speaks to its reputation for consistently delivering high-quality content that champions gender equality. So, while growth is an important strategic goal, Ms. Hyslop says it was important to be thoughtful about how exactly they'd use this technology.
'We can more heavily focus on using AI [on] tasks that are high-effort and low-impact,' Ms. Hyslop explains. 'That's where we're encouraging our team to test, learn and try things that work in their individual processes and also where we're implementing AI as a business. On the other hand, we're not involving AI in low-effort and high-impact tasks.'
For example, the company uses AI to analyze data around audience behaviour so they can better understand what's landing with their readers and listeners and even identify areas of opportunity for new content. Meanwhile, on the tech side, the team is using AI to write one-off codes that can help the company organize all of its data, as well as help sync data between different platforms.
'We use [a number of different programs] and we have to use different scripts to ensure that all of the data from all those different places is actually speaking to each other,' she says.
This approach is exactly the one recommended by Biren Agnihotri, chief technology officer at EY Canada and lead for the company's AI practice, whose role gives him sightlines into a broad range of companies that are integrating AI into their operations.
'Companies succeed when they narrowly focus AI on high-friction, repeatable tasks that require language generation or summarization, not broad, uncontrolled automation,' he says. 'The winning pattern is targeted use case, human oversight and clear ROI (return on investment).'
This holds true regardless of sector or size. Mr. Agnihotri cites examples from across industries, including a financial services company that leveraged generative AI to create regulatory reporting drafts and risk summaries, a SaaS company that 'tied AI to their product roadmap and embedded it in their SaaS platform for customer-facing value,' and a manufacturing company that uses AI for parts classification, supplier queries and maintenance logs.
That's not to say there aren't real challenges to implementing AI, notes Kristina McElheran, assistant professor of strategic management at the University of Toronto Scarborough and the Rotman School of Management.
'Mid-market firms are stuck in the 'messy middle,' where they lack the economies of scale that make AI adoption – and the adjustment costs it entails – worthwhile. At the same time, they are bigger and harder to transform than smaller, more nimble firms,' she says. 'So, for these firms, a lot is going to hinge on how they manage the costs that AI adoption entails. Those that have strong change management skills and the right oversight practices in place may actually do better than smaller firms that lack the dedicated resources for this. It will be hard to match the really big firms, however, for the benefits of sheer scale when it comes to AI. The leading edge of adoption for a long time has been among the largest firms and I don't think that is an accident.'
Cost isn't the only factor. A lack of buy-in from executive leadership, the absence of dedicated compliance and legal teams, volatility in the AI vendor ecosystem and cybersecurity and IP risks can also be serious obstacles. But, perhaps, the biggest challenge for any business is uncertainty.
'Uncertainty around AI isn't just a technical concern, it's a strategic risk,' Mr. Agnihotri says. 'We've seen that [small and mid-size businesses] can't afford to 'experiment and fail fast' the way large enterprises can. As a result, many stall at the pilot stage, unsure whether to commit further investment or scale back.'
The best way to combat this ambiguity is to get very practical. Instead of setting lofty goals around AI transformation, identify the highest-friction tasks in your organization, such as creating reports, summarizing meetings, generating customer emails or responding to FAQs. Determine how you'd measure success (time saved, errors reduced or even how much output increased), then do so via a manageable, replicable pilot project. You likely don't need a custom learning model or proprietary software; instead, lean on the AI tools that are embedded in the programs your company already uses.
And don't underestimate the importance of centring your staff. This means training employees on how to prompt, validate results and recognize when not to use AI; ensuring that your new workflows have people validating, editing or approving results; and, most importantly, creating an AI policy that covers data usage, workflows and prohibited tasks.
'Even a light-touch governance model is better than none and it signals responsibility and reduces future cleanup,' Mr. Agnihotri says.
At The Gist, these conversations have always been part of the process of implementing AI – and they'll continue to be part of future AI adoption plans.
'Here's the thing: change is hard,' Ms. Hyslop says. '[But] change management is nothing new to businesses and change management is something that all founders, business operators and business leaders need to understand. Our leadership and tech teams [understand] AI is another change management thing that we need to be thinking about – in the same way that tariffs are, in the same way that the economy is.'