Laying the foundations for a digital-first future for British Columbia Energy Regulator (BCER)
View client story
View all insights
Seven years to an overnight success: what Alberta can teach governments about AI
Alberta's provincial government recently used AI to turn a $54 million, four-year procurement into a ten-month project at a fraction of the cost. As Minister of Technology and Innovation, Nate Glubish, told us in a recent interview, Alberta's success isn't down to a breakthrough model, but seven years of investment in leadership, trust and capability-building.
Governments around the world are racing to understand what AI means for public services. The Government of Alberta's PRISM initiative has become one of Canada's most compelling examples of what's possible, turning what was expected to be a four-year, $54 million procurement to replace two legacy systems in the Ministry of Infrastructure into a ten-month project delivered by internal teams for around five per cent of the original cost.
Much of the conversation about AI and public services focuses on models and tooling. But after sitting down with Alberta's Minister of Technology and Innovation, Nate Glubish, I was struck by the fact that PRISM is less a story about technology than about leadership, organizational capability and the deliberate work of creating the conditions for AI to succeed.
Seven years in the making
Glubish and I spoke while the Alberta Legislature was sitting. Part way through our conversation, he apologized that he needed to head into the chamber for a vote, and returned a few minutes later to pick up where we had left off. It was an apt reminder that we weren't discussing a completed case study from the comfort of hindsight. We were talking in the middle of governing.
“PRISM might look like an overnight success story, but it took seven years.”
Glubish returned to this point several times, and it reveals the core lesson other governments can take from PRISM. AI has expanded what is achievable, often at a pace and cost that would have been difficult to imagine only a few years ago.
But Glubish’s explanation for PRISM's success focuses less on AI than on what came before it: years of political support, investment in internal digital capability, a culture that encouraged experimentation, and which gave public servants the confidence and permission to work differently.
AI changed what was possible. The seven years before it determined whether Alberta was ready to take advantage.
Leadership creates the conditions
One of the most striking things about Glubish's account of PRISM was what he didn't claim responsibility for. He credits Premier Danielle Smith with providing a clear and consistent mandate to pursue AI-enabled transformation.
“She understands the power and the possibilities of using AI, and she has given me a lot of room to run... That alignment is really important.”
Political sponsorship doesn't deliver a service. But it can give an organization permission to invest, experiment and stick with a different way of working long enough to produce results. Glubish described his own role - of setting direction, providing cover, and securing resources - through a private-sector analogy:
“The CEO raises money from investors... That's my role here in government... My deputy is like my COO... I've brought the resources and the vision. He's building that dream team of talent.”
Glubish consistently puts the credit for PRISM with the public servants responsible for delivery: who built the capability, learned new skills, experimented with AI and delivered the results.
That matters because much of the public conversation about AI focuses on automation, efficiency and what governments might do with fewer people. Alberta presents a different model. Political leaders created the mandate, government invested in its own capability, and internal teams were trusted to experiment. AI allowed those teams to do much more.
Capability before technology
Leadership can create the conditions for transformation, but capability determines what happens next. Alberta spent years investing in its public servants and helping them develop the skills and confidence to use new technologies responsibly. More recently, that has included an open-access Alberta AI Academy alongside clear governance, privacy and data frameworks.
“We build the tools, we build the training, we ensure everyone knows what the guardrails are. And then they can just get to work...”
Governance and innovation are often treated as competing priorities. Alberta's approach suggests the opposite. Clear guardrails gave teams confidence about where they could move quickly, while recognizing and celebrating those who embraced new ways of working inspired others to try it for themselves.
This is a broader definition of AI capability than technical skills alone. It includes public servants who understand the problem, teams that can build and iterate, leaders willing to give them room to work, and the governance needed to do it responsibly.
None of those capabilities arrived with generative AI. Governments have been using user-centred design, multidisciplinary delivery and rapid feedback loops to deliver better digital services for years. Alberta demonstrates what can happen when AI arrives in an organization that has already invested in those capabilities.
Building confidence to experiment
Governments have good reasons to be cautious about failure. But avoiding failure entirely makes it difficult to learn. Glubish described adopting a portfolio approach more commonly associated with the private sector:
“You need to have several initiatives... We've had these wins over here like PRISM... We've had a few over here that didn't go anywhere. That's normal. That's how we learn.”
During PRISM, the team developed hundreds of prototypes. Most were never deployed, but they reduced uncertainty, informed decisions and helped the team find better solutions.
“Give me someone who's entrepreneurial and willing to take a risk... over someone who says, 'I'm just not going to rock the boat.'”
It is a striking comment to hear from a minister. AI can make experimentation cheaper and faster, but that only matters if people have meaningfully been given permission to experiment in the first place.
AI changes the build-or-buy calculation
The headline figures behind PRISM are extraordinary. A project expected to require a four-year procurement worth more than $54 million became a ten-month delivery led by internal teams for around five per cent of the original cost.
For decades, governments have often approached large digital transformations through major procurements and long implementation timelines. AI is beginning to challenge some of the assumptions behind that model, empowering relatively small, capable internal teams to achieve things that would previously have required much more time and money.
That strengthens the case for investing in internal digital capability. It should also cause governments to reconsider work they may previously have outsourced.
One project isn't enough to declare the old model dead, nor will every government have the capability to build everything itself. But PRISM should prompt some harder questions. What capabilities should governments regard as core? When should they build rather than buy? What happens to the business case for a large, multi-year procurement when an internal team equipped with AI can prototype and deliver at a fraction of the cost?
There are paradoxes here. AI is sometimes presented as a way for organizations to function with less internal capability, but PRISM shows that when AI allows capable public servants to achieve much more, the value of having those people and capabilities inside government increases.
At the same time, governments are often drawn to automation in the hope of cutting costs. Alberta's story shows that the real savings come not from replacing your team, but from skilling and empowering it to take on work that would otherwise be outsourced.
A fresh perspective on AI
There is understandable excitement about what AI tools might automate in the public sector, and the pace at which it might be able to transform services.
Alberta offers another perspective.
Glubish spoke less about replacing public servants than about inspiring and enabling them, and less about overnight tech-enabled transformation than about the slow, deliberate work of building the right conditions. Political leadership created the mandate. Years of investment built the capability. Internal teams were trusted to experiment and deliver. AI dramatically increased what those teams could achieve.
Alberta’s story shows that the governments likely to benefit most in the intelligence era won’t be those who chase quick wins, or the newest models. They will be the ones that have spent years building the internal skills, culture, and ways of working so that AI can enhance - not replace - their capabilities.
Written by
Alexandra MacEachern
Director