When an organization hands generative AI to every employee, the rollout is rarely the interesting part. What happens a year later is. ATB Financial ran that experiment in public, and one of its executives has since been unusually candid about the result.

In January 2025, Google Cloud announced that ATB had deployed Gemini for Google Workspace to more than 5,000 team members, inside the tools they already lived in, making it one of the first major financial institutions in Canada to do so. The rollout followed a 2024 pilot involving several hundred employees that produced exactly the kind of numbers that get a budget approved. Participants reported average time savings of up to two hours a week. Sixty per cent said they were able to accomplish more in the same amount of time, or take on additional responsibilities.

The deployment worked

By any conventional measure this was a success. Adoption was quick, the tooling was mainstream rather than experimental, and the deployment itself won a 2025 CIO Award from IDC and CIO Awards Canada, recognising both the organization-wide rollout and the governance framework around it. Over the same period ATB has been repeatedly recognised as one of Canada's best workplaces, ranking fourth nationally in 2025 and first in its size category in 2026. Nobody involved needs to be defensive about the decision.

The sentence worth reading twice

Speaking to Human Resources Director in June 2026, Tara Lockyer, ATB's Chief People, Culture, Brand and Communications Officer, said something most organizations would never put on the record: "We've definitely seen huge individual productivity uptake, but we're having a really hard time harnessing that for the organization."

She was equally direct about the rollout itself. "We didn't necessarily set an expectation in terms of usage," she said, adding that she would establish clear usage benchmarks from the start if she were doing it again.

Read those two statements together and, in my view, you have the most useful applied-AI lesson any Canadian institution has put on the record this year.

Two hours a week is not a business case

The temptation is to multiply. Two hours by five thousand people is ten thousand hours a week, and on a slide that is an enormous number. It is also not a number anyone measured. The two hours came from a pilot cohort of a few hundred self-selected participants reporting on themselves, which is a reasonable indicator and a poor forecast.

Even if it held, it would be ten thousand hours scattered across five thousand calendars in fifteen-minute fragments, and fragments do not aggregate on their own.

Saved time becomes organizational value only when something deliberate happens to it. A process is redesigned around the shorter cycle. A backlog is cleared. A role is rebalanced toward work that was previously deferred. A service level is raised. Absent that, the time is genuinely saved and genuinely invisible: employees feel the relief, and the income statement never hears about it.

Why capable organizations land here

This is not a failure of technology, nor of effort. It is a structural mismatch between two decisions that get made very differently.

What to do differently

Each of those is easier to write down than to implement. I have written a companion piece on what has to change inside the organization for the saving to become visible: Capturing the Saving: What Has to Change After the AI Lands.

Why this case is worth studying

Most cautionary AI stories involve a technology that underperformed, which makes them easy to dismiss: wrong vendor, wrong model, wrong moment. ATB's does not. The tool did what it promised, people used it, and the productivity was real enough that the deployment collected an award. That is precisely what makes it instructive, because it isolates the variable. When the technology performs and the value still does not land, the constraint was never the technology.

Execution, not technology, is where transformation succeeds or fails. ATB has been candid enough to show the rest of us exactly where the seam is, which is more than most organizations are willing to do.

Sources: Google Cloud press release, 23 January 2025; Human Resources Director, 11 June 2026; ATB Financial fiscal 2025 and fiscal 2026 results releases.