From Kelowna to Selkirk: How Canadian Municipalities Are Quietly Leading in AI

It's 11:47 PM on a Tuesday in Selkirk, Manitoba. A resident has a question about a water service disruption. City Hall closed hours ago. There's no after-hours staff answering phones.

But someone picks up. Her name is Charlotte.

Charlotte is an AI phone agent deployed by the City of Selkirk in January 2025. She answers calls after hours, documents service requests, and submits tickets during the conversation. She doesn't replace staff — she extends the city's reach into the hours when residents still need help but nobody's at the desk.

Selkirk isn't Toronto. It isn't Ottawa. It's a small city in Manitoba that decided AI wasn't just for big municipalities with big budgets. And it's not alone.

The Numbers Behind the Shift

According to the MNP/Leger 2025 Municipal Report — the most comprehensive annual survey of Canadian municipal technology leaders — 23 percent of Canadian municipalities are currently using AI. Over 50 percent are actively exploring or planning adoption.

The most striking number: AI as a stated priority jumped to 48 percent in 2025, more than doubling from the prior year.

This isn't hype-driven. The top municipal priorities tell the story of what's actually driving decisions: cybersecurity and privacy (78 percent), citizen experience and service delivery (65 percent), technology modernization (63 percent), AI (48 percent), and smart cities and digital infrastructure (47 percent, up 12 percent from 2023).

AI adoption is climbing because it sits at the intersection of the things municipalities already care about — better service delivery, operational efficiency, and doing more with constrained budgets.

What Early Adopters Are Actually Doing

The Canadian municipalities leading in AI aren't deploying experimental moonshots. They're solving specific, practical problems.

Kelowna, BC became the first municipality in Canada to launch an AI-powered building permit digital assistant in October 2023. In its first two months, it handled over 9,500 questions. Eighteen percent of those interactions happened outside business hours — residents getting answers when staff weren't available, without adding a single overtime hour.

Winnipeg, MB launched an AI chatbot pilot in May 2025 that supports 17 languages and runs 24/7. For a city with significant linguistic diversity, this isn't a nice-to-have — it's an accessibility tool that makes municipal services navigable for residents who might otherwise struggle with English-only phone menus.

Markham, ON introduced "Millie" in April 2025 — Canada's first municipal mascot paired with an AI chatbot. But Markham's AI journey started earlier: during COVID-19, it became the first Canadian municipality to deploy IBM Watson Assistant for Citizens, handling a surge of pandemic-related inquiries that would have overwhelmed call centres.

Burlington, ON launched "CoBy" in February 2024, one of the first Canadian municipal generative AI assistants built on Microsoft Copilot Studio. It helps residents navigate city services through natural conversation rather than menu-driven navigation.

Small Towns Are Moving Too

The assumption that AI requires a large IT department and a substantial budget is being disproven across the country.

Selkirk, MB deployed both a website chatbot ("Chuck") and the AI phone agent ("Charlotte") with a municipal team that's a fraction of what larger cities employ. Charlotte's after-hours phone capability is particularly notable — it addresses a genuine service gap without requiring additional staffing.

Milton, ON deployed a phone-based AI agent specifically for winter service updates in January 2026. During a major snow event, it handled 435 incoming calls, resolved nearly half independently, and managed 297 minutes of call time. That's roughly five hours of phone conversations handled without a single staff member being pulled from operational work during a storm response.

Barrie, ON launched an AI chatbot pilot on its website in May 2025 — straightforward, practical, and available 24/7.

These aren't prestige projects. They're municipalities addressing real capacity constraints with tools that are increasingly accessible and affordable.

The Gap Between Interest and Implementation

Despite the momentum, the MNP data reveals a significant readiness gap. Nearly one-third of municipalities still lack formal AI guidelines or policies. Nearly half still rely primarily on Excel for analytics.

The top barriers municipalities report are telling: insufficient resources (50 percent), complicated decision-making processes and outdated procedures (44 percent — double from 2023), legacy technology systems (43 percent), and lack of subject matter expertise (36 percent).

This means the challenge isn't convincing municipal leaders that AI matters. That argument is largely won. The challenge is helping them move from interest to implementation — navigating procurement, addressing governance gaps, managing change internally, and finding solutions that work within their specific constraints.

York Region recognized this and created a General Manager of Digital, Data, Technology, and Customer Experience — a role designed to unify functions that traditionally operate in silos. King Township published a comprehensive Digital Transformation Framework with a responsible AI roadmap. These are governance-first approaches that treat AI as an organizational change, not just a technology purchase.

The Economics Are Clear

Perry Group's research quantifies what municipal staff already know intuitively: a face-to-face transaction costs $15 to $30. A phone call costs about $5. An online or AI-assisted interaction costs roughly $1.

When a chatbot handles 9,500 questions in two months (Kelowna) or an AI phone agent manages 435 calls during a snowstorm (Milton), the math speaks for itself. These aren't replacing human judgment — they're handling the routine inquiries that consume disproportionate staff time.

The freed capacity matters more than the direct cost savings. Staff who aren't answering the same questions about garbage collection schedules or office hours can focus on permit reviews, infrastructure planning, bylaw compliance, and the complex work that actually requires expertise.

What Comes Next

The municipalities featured here aren't waiting for a national AI strategy or a provincial directive. They're identifying specific pain points — after-hours accessibility, multilingual service, high-volume routine inquiries, storm response communications — and deploying targeted solutions.

The pattern is consistent: start with a defined use case, measure the results, and expand based on evidence.

For municipalities still in the exploration phase, the playbook from early adopters is clear. Build governance first — AI policies, data management frameworks, responsible use guidelines. Start with high-volume, low-complexity interactions where the ROI is easiest to measure. Choose tools designed for municipal use cases, not generic enterprise platforms that require months of customization.

Platforms like Muni are built specifically for this context — helping municipalities deploy AI-powered resident services that work within the realities of public-sector governance, budgets, and accountability.

The tipping point isn't coming. For Canadian municipalities, it's already here. The question isn't whether to adopt AI — it's how quickly you can move from interest to implementation before the gap between your municipality and its peers becomes a gap in service quality for your residents.