Ottawa's New Gatekeeper Won't Take Meetings
Government relations has always been a human business. In Ottawa, nothing works quite the way people think it does. The org chart is a polite fiction. Power moves through invisible networks of overworked staffers, assistant deputy ministers, and policy advisors whose names you’ll never read in the paper. These are the so-called nobodies who actually run the country.
That was the argument in a recent post of mine. It still holds. But it may not hold for long.
We’re on the edge of something new. The next evolution of government decision making won’t be led by an AI supreme intelligence. But it may be guided by something that looks like one. Or at least something that speaks with certainty and comes wrapped in evidence.
Models are only as good as their assumptions. Once those assumptions start guiding decisions, they matter a lot.
Last year, the Tony Blair Institute published a report on the future of government in the age of artificial intelligence. The whole thing is fascinating, but one idea stuck with me more than any other: a National Policy Twin by 2030.
The National Policy Twin is a digital replica of the machinery of government. It doesn’t just simulate a single policy. It integrates real-time data from across departments and builds a dynamic model of the country itself from tax and spending flows to demographic changes and public service delivery. It is, in effect, a virtual Canada. A simulated test bed where new ideas can be trialled without political risk.
Raise a tax. Cut a benefit. Introduce a new credit. The model shows what happens to the economy, the population, and the budget. Ministers won’t just be briefed. They’ll be shown outcomes before they occur. Not what if, but here’s what.
It’s not hard to imagine how that changes things for lobbying.
Because you can’t lobby a simulation the way you lobby a person.
The model doesn’t care how long you’ve been on the file. It doesn’t care if your CEO is in town. It doesn’t care how many times you’ve met with the department. If your issue’s not in the data, if it’s not built into the assumptions, if it can’t be seen in the model, then it won’t show up in the answer.
You’re not late to the conversation. You’re not in it.
This isn’t science fiction. Some governments are already experimenting with early versions of this approach. And now Canada’s positioning itself to follow suit.
Prime Minister Mark Carney’s recent mandate letter to Cabinet make it clear. Government must become more productive by deploying AI at scale, focusing on results over spending, and using scarce tax dollars to catalyse private investment. The appointment of Evan Solomon as Minister of Artificial Intelligence and Digital Innovation shows this isn’t just about service delivery. It’s about how government governs.
If that future arrives, the old lobby playbook won’t work the way it used to.
We’ve seen shifts like this before. The first came when corporate and union donations were banned in the early 2000s. That ended the days of buying a ticket to the minister’s golf tournament. The donor circuit vanished. Firms had to sharpen their value and compete on strategy and narrative, not just access.
The second came with the Federal Accountability Act. The Harper government introduced stricter post employment restrictions. Staffers had to sit out before entering lobbying roles. That slowed the revolving door and made relationships rarer. Government relations got more professional. More structure, less handshakes.
Now comes a third shift: algorithmic policymaking.
This is still a human business. Ministers will still make the decisions. Deputies will still advise. Chiefs of staff will still manage the politics. But those decisions will lean more and more on data and predictive modeling. That foundation may not be questioned. It’ll be seen as neutral. Objective. Evidence based. Even when it’s not.
To be fair, the best GR pros have always operated upstream. They’ve helped shape how problems are defined, what counts as evidence, and which trade offs are worth making. What’s changing is that those same decisions are increasingly being filtered through models that present themselves as neutral and data-driven, even when they might not be. The rules of influence will shift from conversations to configurations.
Picture a housing nonprofit proposing a new rent supplement. Under the old model, they’d build political support and pitch a pilot. In the new model, the department runs a simulation and says it doesn’t move the affordability needle enough. The proposal dies in the model, not in the minister’s office.
Before anyone writes an obituary for the profession, let’s be clear. The model might predict what’ll happen if you change a policy. But it can’t tell you what ought to happen. That still belongs to politics. To values. To elected people deciding what kind of country they want to build. The Policy Twin might help navigate the road. Politicians still choose the destination.
Lobbying isn’t going away. It’s just shifting again. The job’s still about helping shape those choices. Explaining trade offs. Showing what matters and why. Making sure your client’s priorities are not just in the data, but part of the values conversation too.
But it does mean lobbyists will need new tools. That’s a different skill set. It means knowing how departments collect and structure their information. Knowing what proxies they rely on. Helping clients frame their interests in ways a model can ingest and understand. It also means explaining failure. Sometimes the answer won’t be political. It’ll be computational.
There will be fights over how these models are built. What gets counted in the algorithm and what gets left out. These are political choices, not technical ones. And as models become central to policy design, GR professionals may need to build better links with the people who feed them. That means knowing how to engage not just with departments, but with the academics, think tanks, and researchers whose work often becomes the assumptions inside the machine.
The fights over how the algorithm is designed will be real. GR professionals will be in the room for those fights. And increasingly, they may need to be just down the hall from the data scientists and policy modellers too.
The model will become the meeting. If you want your issue on the table, it’ll need to be in the simulation.Ottawa’s next gatekeeper may not be a person. It may be a model.
If done right, this shift could make policymaking more transparent, more accountable, and more rigorous. But only if we remember what the model can’t see. Public trust. Lived experience. Political will.
I’m relieved at this development. I’m a horrible golfer anyway.