Canada’s General Fusion is now the first publicly traded nuclear fusion company, as its Nasdaq debut gives it more cash to chase a breakthrough that could one day help meet soaring demand for clean power — if the still-unproven tech can finally produce more energy than it uses. Plus, doctors are warning about a tick-borne illness that was barely on their radar five years ago, as warming temperatures help ticks spread farther north and bring more diseases with them.
And in The Big Picture: Ottawa and Alberta strike a deal with oil sands producers on a major carbon capture project, an Innu First Nation rejects Quebec’s proposed hydro deal, and Chinese automaker Dongfeng gets ready to sell EVs in Canada.
After today's interview we host an exclusive interview with Sumee Seetharaman, VP of AI/ML Practice in the Center of Excellence at TD:
Brett:
Hey, Peak Pals. Thanks so much for joining us today. We've got a great guest for you, Sumee Seetharaman, who is the VP of AI/ML Practice in the Center of Excellence at TD.
This is gonna be a great conversation about AI at work, something I think we're all experiencing, so we're really excited to have you. To kick it off, the use of AI in Canadian workspaces is surging. According to KPMG, about 51% of Canadian adults are now using generative AI at work. Has the technology become just so commonplace in the workplace that Canadians need to have AI skills before they enter the workforce? And if so, which skills do they need?
Sumee:
Yeah, it's a great question. Interestingly, we're at a point where the personal use of AI is far exceeding use at work, a March 2026 Angus Reid survey found that over 70% of Canadians surveyed had used AI in the previous three months for personal use. So odds are that most folks are actively using AI for their day-to-day already and have some of the basic AI skills required at work. Skills like knowing how to use and prompt a chatbot effectively, and I'm not talking about just a single Q&A, but really knowing how to converse with the model. Treating, the response to your initial question as just the starting point. Knowing how to ask follow-up questions and providing clear instructions to refine the response. So the fluency is in really knowing how to iterate and refine, building out the parts that are missing, and pruning out the parts that don't feel right? Most people are learning to do this instinctively as they're using ChatGPT and Gemini apps.
Next, I want to say folks who are using AI regularly are also starting to develop a good sense of the strengths and the limitations of the technology, right? All of us have encountered that the model sometimes hallucinates, and so we know and we've learned that we cannot treat the output of the model as gospel. And so in a work setting, this is where domain expertise, not AI expertise, is actually more helpful because we need experts to validate and apply human judgment on the output of the model. And also if you're actively using AI and reading on AI, AI can be used in harmful ways, so most folks are starting to develop an appreciation for the guardrails that are needed and what does responsible use of AI look like. And are careful about what you put into public versions of these models. You're starting to build a good awareness of privacy safeguards and such. All that to say if you're familiar with using ChatGPT or Gemini for personal use- You should be able to fluently use most AI tools in the workplace. So at TD, over seventy percent of our colleagues our workforce has access to a GenAI capability, whether it's virtual assistants that we have developed in-house for our frontline colleagues in contact centers or wealth operations or branches, or Copilot capabilities we've enabled across the organization. GitHub Copilot for developers and Power BI Copilot and Office 365 Copilot. So chances are, if you're joining TD, you have access to these AI capabilities to, to really amplify your impact, and you probably are already quite fluent in how to use them.
Brett:
TD is obviously ahead of the curve on AI, and I'd be interested from a practical standpoint, what are some examples of jobs at TD that AI is already transforming?
Sumee:
Yeah. So we recently announced our first agentic AI use case to, to speed up the pre-adjudication process of mortgage and HELOC applications. So when an application is submitted, AI agents that we've deployed are actively scanning the documents, calculating client income, performing consent checks, looking for discrepancies, et cetera. After this sort of document review process is done, the agents then produce a memo with their findings and this goes off to the underwriters to review and make a decision on the application. And so if you look through that agentic AI activation, it's transformed the jobs of colleagues like our underwriters who are now working hand-in-hand with AI agents and making decisions on applications much faster. We've gotten the review process down to an average of three minutes from an average of fifteen hours. Effectively, the way they're operating has transformed day-to-day. If I look at the role of our back-office staff that has transformed as well. They're now providing valuable oversight on our AI agents and helping us ensure that it's performing within the necessary parameters and it's delivering the right outputs. It's just the first set of use cases supporting the first pre-adjudication leg, but there's a much broader end-to-end transformation roadmap for our real estate secure lending operations. So we're just getting started. And as we gradually scale AI use within RESL operations, we'll start to see more role transformation across the board. On a similar vein I find the roles of our frontline colleagues are gradually evolving, transforming too. So our colleagues in branches are now using gen AI virtual assistants to help prepare for client meetings, answer complex questions effectively. Our colleagues in TD Securities are now using these virtual assistants to speed up the research process and really amplify the sales cycle. So across our frontline, our new hires now have an expert AI support on their fingertips, which means they can now spend far more time focusing on sales advice, deepening relationships with our clients. So slowly but surely, we see these roles evolving, getting actually much, much more client-focused and less manual work and less operational focus, if you will.
Brett:
What is TD doing to help train and upskill colleagues so that they can build the AI skills they need to thrive in these transformed roles? And I guess how do upskilling efforts differ for an individual who's a beginner with AI versus a new hire at Layer 6 who's looking to build their AI skills?
Sumee:
Yeah, it's a great question. We're actually employing a multi-pronged training strategy for AI across the organization. Obviously it's customized to your function, your role within the organization, but it takes many formats. It could be trainings, recordings, live, virtual or in-person immersive hands-on sessions, hackathons, longer-term programs and pilots, et cetera. Across the gamut, depending on the role you're in, depending upon the function you're in. I'll maybe talk about a few initiatives, what we've rolled out across the organization. The first one I'll highlight is TD Thrive, which is our enterprise learning platform. It has dozens of AI courses that are available on demand to every TD colleague, and these cover a broad spectrum of topics from one-on-one sessions like getting started on LLMs to very advanced, architecture of agentic systems, as an example. Nearly 4,000 colleagues have completed the AI Foundations Pathway on TD Thrive, a program designed to build colleagues' understanding of AI and introduce our Responsible AI approach. Another 8,138 are currently enrolled in the program and on pace to complete it.
At the executive level, we have rolled out through our strategic academic partnerships, we have rolled out executive AI education. Last year alone, almost five hundred TD leaders attended this two-day AI training course led by business school profs from Columbia University. VizMasters is actually a hackathon we have run for six years in a row. This year, we had about three hundred people from across seventy-five teams participate. And so this year the contestants used things like Power BI Copilot, which allowed them to turn data quickly into charts and dashboards delivering business insights through simple natural language prompts. And we found as a result of these hackathons, the usage of Power BI Copilot actually surged by about two hundred percent during the competition. So you're engaging, you're learning the skill set, and then you're starting to adopt and use it to supercharge your work. Another example I'll quote is DataCamp. This is on the longer-term format. This is actually a six-month pilot we recently launched with our analytics and AI colleagues which really has a hundred and sixty-five plus AI ML courses, nearly sixty hands-on AI projects. And we observed that just in the first two weeks of rollout, almost half of our organization had registered, logged over, five hundred hours of learning, which has been fantastic to see. To your question on AI beginners across the organization versus Layer 6 we roll out these training initiatives programs broadly for colleagues within the bank, but colleagues who are in Layer 6 are already expected to operate at the forefront of AI. And so they typically attend training, learn through joint research with our academic partners, whether it's Columbia or MIT or UofT. It's almost a role expectation that they continuously upskill and operate at the bleeding edge of AI. That's really neat, and it's-- these sound like really great and effective initiatives to upskill your workforce.
Brett:
How do you know that upskilling is actually working and that your colleagues are actually adopting AI skills and using them in their day-to-day workflows?
Sumee:
Yeah. Every time we develop and roll out these capabilities, whether it's in-house or it's commoditized capabilities and we roll it out to colleagues in branches or contact centers, what have you, we have instrumentation mechanism in place to track, monitor usage and efficacy of the solutions. So as an example, when we rolled out our branch virtual assistant to roughly eighteen thousand colleagues, branch colleagues, we immediately saw a near seventy-three percent adoption rate of the capability. And these are-- these day two signals are super valuable because they tell us a few things. They tell us our training change management initiatives are successful. They tell us that the model is effective and useful in the business setting. It's also good feedback on what's not working and where we need to iterate and improve our product to make it more relevant to colleagues. Similarly with Microsoft Office Copilot, our adoption rate is currently at around eighty percent. So having those clear metrics to measure, quantify use day two is super important to really realize the value of AI, and it tells us that our upskilling training initiatives are working and colleagues are actively using these capabilities. For more senior audiences, we look for a different set of signals to gauge if, executive training is effective and it's driving the desired outcomes. So these initiatives are successful when our leaders are equipped to drive AI transformation within their businesses, can really help their organizations transition from, one-off use case thinking to more three-sixty transformational roadmaps.
Brett:
That is a really great way to track whether or not this upskilling is actually working. And I think what you're probably seeing here is, I, and I know you've described it as this before, but that a rising tide lifts, that li-lifts all boats. I know it's really hard to forecast knowing how fast technology's changing. But looking out a few years, how do you think AI changes what it means to work at a bank?
Sumee:
Yeah, I'd love to have a crystal ball. The AI wave is often compared to scaling and adoption of electricity or the internet. And so really if I contrast to that, then a few decades from now, we wouldn't be talking about AI as a differentiator. It'll just be in the fabric and the wiring of how our day-to-day processes and technologies work. It'll be invisible. And so as we think about driving AI adoption at TD we're really taking a hybrid approach. We see AI capabilities amplifying colleague impact to ultimately deliver better results for our clients. And you see this kind of approach play out in the RESL example I was sharing earlier, where the AI agents are deployed, they're handling document reviews, but ultimately it's our colleagues who are making-- applying their expertise, making decisions on applications. So we're going to continue to drive these kinds of workflows and agentic systems across the bank where AI takes on a bigger role optimizing automating manual tasks. But colleagues with specialized skill sets drive domain expertise and provide oversight in these workflows to make informed decisions. So really, at the end of the day, we feel colleagues that have the specialized expertise that apply their judgment to drive clear outcomes interacting with AI will, will-- is the target state for us. We expect our colleagues to be AI fluent and proficient so they can leverage AI to drive outsized impact. So naturally, folks with a learning mindset, a sense of curiosity, ability to actively learn and harness AI effectively will naturally thrive. That's true in Layer 6, and it's just as true anywhere else within the organization.
Brett:
Thank you so much, Sumee, for joining us today. It was great just to learn about how AI-- how TD is deploying AI, both at a kind of high level but also at a practical level. And so thank you again, Sumi. It sounds like there's a lot going on at TD, and so we'll let you get back to your day. But I appreciate you taking the time and joining us today.