"How freaked out do you want to be?"

That was Salim Ismail's answer when someone in the audience at KPMG's AI Made Real summit in Toronto last week asked whether government policy could keep AI in line. 

Needless to say, he's not optimistic. Not what you want to hear from a keynote speaker who just spent an hour explaining that AI is improving so fast he thinks most organizations won't survive the next decade in their current form.

Ismail is a serial entrepreneur, the founding executive director of Singularity University, and the author of Exponential Organizations. He has spent the better part of two decades making this case to Fortune 500 CEOs, governments, and once, the Vatican. 

Read more about his keynote and what the panel that followed had to say here. 

When we cover an event like this, there's always something that doesn't quite make it into the published article. 

But you are a newsletter subscriber, so more on that below.

Ismail's argument hinges on what he describes as a company's immune system. The instinct, in any organization built around something that works, to protect what's working and kill off the thing that might replace it. 

Promising ideas die in the budget process, the legal review, the meetings that don't get scheduled. The body is doing what it's supposed to do. In this case, it's doing it to the future.

He offered Marriott as an example. The company had every idea inside it that became TripAdvisor, Airbnb, and Booking.com, he said. None was built. The hotel business was working too well to risk. Had Marriott spun those companies out, Ismail said, it would have a market capitalization three times its current size.

For a counter-example (the one that didn't make the article), he pointed to a Chinese used-car company.

The business model itself is almost embarrassingly simple. 

They price a car at what they think it's worth, offer the seller 10% below in cash on the spot, and sell to the next buyer for 10% above. 

Not rocket science, but what makes it scale is a quick, consistent method for determining the car's value. Photos and engine audio get run through a machine learning model that returns an accurate price. 

They launched in 2017. Seven years later, they had 80% of the used-car market in China.

Without an immune system to fight, they could build a business model around AI from day one and scale it fast.

A similar dynamic is evident in my colleague Jennifer Kervin’s article about why Canadian technology leaders keep buying from foreign vendors even when they say they'd rather not. 

The risk-averse board, the slow process, the comfort of the recognized name, all of it pulls the contract back to the incumbent. The system that protects what's worked before is the same system that blocks what could come next. 

Government grants are a common policy lever for backing Canadian innovators, but Kervin's reporting points to a more powerful one. 

"A purchase order is incredibly more valuable than a grant," says Patrick Searle, vice president of policy and government affairs at the Council of Canadian Innovators.

Some light reading for when your company's immune system is winning a few too many fights.

Canada's patent picture reveals a commercialization gap. Canada has the inventions, but too many of them sit with universities and government agencies instead of companies that can turn them into products. In biotechnology, only one of the country's top 10 patent holders is a private firm. In the U.S., that number is six. 

CGI and Schulich open a new path into enterprise IT, with no coding experience required. Canada ranks 44th out of 47 countries for AI training and literacy. This program aims to grow the talent pool by training new people from scratch, so companies have somewhere to hire from beyond each other.

Visa brings its Agentic Ready program to Canada. All five major Canadian banks are early partners. Agentic commerce, where AI agents initiate purchases and payments without direct human approval, is moving from concept to live infrastructure faster than the governance frameworks meant to handle it.

For the most part, I'd describe myself as a techno-optimist. The tools are getting better, the costs are dropping, and the people building with them are mostly trying to do something useful. 

But there are days, and certain keynotes, that leave me a little uneasy. The technology will do what it's built to do. What I'm less confident about is whether the humans holding it will use it responsibly and whether we have the political leadership to manage societal changes that will come with it. 

So how freaked am I? Medium freaked out, these days.

Next week, Digital Journal will be in Edmonton for Upper Bound, Amii's annual AI conference, where Digital Journal is the official media partner. 8,000 attendees, sold out. If you’ll be there, make sure to say hi. 

— David