How this Ont. town is using AI from bylaw to trash pickup

King Township, a small municipality in Ontario's York Region, is deploying artificial intelligence across its operations, from waste management analytics to parking tickets, positioning itself as one of the more active early adopters of the technology among Canadian local governments.

Marco Cheng-Perri, who joined the township in 2024 in the newly created role, Manager of Digital Transformation, Data and Corporate Performance, has been leading the initiative.

His background includes work on the City of Markham's AI-powered chatbots during the COVID-19 pandemic and a subsequent period at Amazon. King Township's Chief Administrative Officer brought him in specifically to drive agentic change within the municipality.

The township's AI work is built on two platforms: Snowflake, used for data aggregation and analytics, and Microsoft, used for general productivity and project renderings.

"When I get an email from a vendor with a bill, I want to be able to take that email and automatically pull the data off and put it into my tracker: Typically, that's going to be better suited for a Microsoft agent because the connection to Outlook is just really strong. But if someone's asking 'we have these three business systems and we want to get deep insights into how things are operating across them,' that's going to be a Snowflake project for sure," said Cheng-Perri.

The challenge that Snowflake addresses is one common to municipal government: data fragmentation. King Township has been using Snowflake for approximately one year, a period that has spanned the platform's introduction and subsequent rebranding of its intelligence layer, now called Snowflake CoWork.

The township operates across dozens of service lines, from bylaw enforcement to fire services, each with its own systems and reporting cycles. Historically, this data has been siloed.

Data sources feeding the system include Google Analytics and waste collection figures provided by contractor GFL. The township is also working to integrate geospatial data through Esri, which Cheng-Perri expects will open up new use cases within public works.

"Building the AI agents and dashboards, that stuff is easy these days. The challenge is actually within the data," added Cheng-Perri.

One of the more technically involved internal projects is an AI agent built to calculate development charges, a process that draws on regional and municipal bylaws, submission data and development timelines.

"I underestimated how complicated these were when I took on the project ... Understanding the timelines for development and getting those final numbers typically takes hours for staff. But to create an AI agent that understands all of these bylaws, rules, and principles that can accept these submissions and proposals, and then spin out a calculation for you in minutes, that's big," added Cheng-Perri. "We're applying this logic to a ton of processes internally to free up staff time to focus on higher value tasks."

On the resident-facing side, the township has deployed a chatbot, "Kingsley," on its website. The tool was deliberately kept narrow in scope at launch, drawing only on website content rather than being connected to live service systems or payment infrastructure. The assistant, launched in June 2024, can execute tasks from providing weather forecasts to explaining traffic studies.

"We could have connected it to our CRM. We could have let people pay their bills through it. But there's too much risk there for us. We're innovative, but we make sure that we're respecting guardrails and privacy. We just loaded it up with website data and that's been tremendously beneficial to a ton of people," added Cheng-Perri.

"We're small at King Township, but we're very proud of that. We're basically trying to show everyone that innovation isn't reserved for large municipalities or large organisations. With the right focus on innovation while also balancing responsibility, you can do a lot of cool things, too."