Newcomer Compass: A cross-government navigator for the benefits and services new Canadians qualify for

This project would develop an application that allows new permanent residents in Canada to easily understand the help they are entitled to.

Currently, new permanent resident struggle to navigate the support available, as the information is scattered across a range of different federal and provincial ministries, and settlement agencies –  each with its own forms, rules, and language.

As a result, newcomers miss support they are owed, and frontline workers can spend hours piecing together what one person qualifies for.

Newcomer Compass would provide a better solution by taking a newcomer’s profile – with their consent – and return a personalized, plain-language map of the benefits and services they qualify for across government, with the next step for each. It would also give caseworkers and settlement workers a single view instead of a dozen tabs, all built on a maintained set of machine-readable eligibility rules matched against a person’s situation.

Submitted by Jad Seaidoun at Immigration, Refugees and Citizenship Canada

Early Flood Warning System for Canadian Infrastructure

Flooding is Canada’s most common and costly natural hazard, causing average annual losses of over $2bn – and the risk is growing

The existing federal flood risk finder lets Canadians check whether an address is in a flood-prone zone, but the map does not provide live information – instead it provides a static map of where flooding is possible, not when the risk is rising.

This early flood warning system would create a model that provides that fills that gap. It will use Environment and Climate Change Canada’s HYDAT river discharge data and daily climate data to flag short-term flood risk at individual monitoring stations across the country. This system would also pick up real seasonal patterns, like spring snowmelt risk, that could cause flooding, creating an operational tool used alongside infrastructure and emergency management teams – and could form a building block of a centralized natural disaster warning system for Canada.

Submitted by Umair Butt at Housing, Infrastructure and Communities Canada 

CanService AI: AI for Service Demand Forecasting

Service delivery departments across the Government of Canada face frequent mismatches between forecast demand for services in departments and agencies across the Government of Canada, and the actual demand.

This disparity does not just have a theoretical impact – it has big knock on effects on service delivery and creates long wait times, uneven workloads, and inefficient use of staff.

CanService AI would solve this problem by using agentic AI and predictive analytics to better forecast service demand across departments and service channels (in-person, digital, call centres). By joining up existing data sets, analyzing historical volumes, seasonality, policy changes, and external factors, the model would generate dynamic, near real-time demand forecasts.

These insights would enable proactive workload management, optimizing staffing, reducing bottlenecks, and improving service timelines.

This initiative enables a proactive, integrated, and AI-enabled public service with a scalable, whole-of-government approach to anticipate demand, improve outcomes, and deliver better services to Canadians.

Submitted by Lucia Goncalves at Employment and Social Development Canada

Reading the Notes We Already Write: Using AI to Spot Food Safety Risks Earlier

Every week, food safety inspectors write down what they see in notes, records of problems, and follow-ups. In these notes are early warning signs of problems – the same issue appearing again and again across different sites, a pattern in a particular product, or small, repeating failures.

However, these notes currently get filed, and rarely read across cases, meaning that a pattern which is obvious when hundreds of reports are looked at together is invisible to any single inspector.

This AI tool would read all the inspection notes already collected and highlight patterns worth a closer look, but which would otherwise be missed: grouped by site, region and product type.

The tool doesn’t make decisions. It raises possibilities, and an inspector checks them. But it would make smarter use of inspector time, and cut the time from a warning being written and someone acting on it from weeks to days – meaning fewer costly recalls.

Submitted by Dreck Niyonsenga at Canadian Food Inspection Agency

Using AI to Support Packaging Compliance in Medical Devices

Navigating regulations on medical devices is complex, with regulatory review of packaging and labels forming a manual, labor-intensive bottleneck to slows helpful medical products getting to market.

The Packaging Compliance Agent is an innovative AI-driven vision and text workflow designed as an internal decision-support “copilot” for public servants. By analyzing text hierarchies, contrast ratios, and bilingual balancing simultaneously, the agent flags potential non-compliance indicators in seconds. This transforms the user experience. Instead of manually auditing tedious checklists, analysts are empowered with an automated advisory tool, allowing them to focus their expertise on high-risk, complex case reviews.

This tool would lead to a drastic reduction in submission backlogs and processing times. Taxpayers and businesses would benefit from accelerated product launch timelines, while departments could optimize resource allocation by shifting highly trained human capital away from repetitive data validation. By keeping data processing tightly scoped, the project exemplifies the responsible, transparent, and ethical deployment of AI.

Submitted by Jovian Tsang at Canadian Institutes of Health Research

Canadian Digital Twin of the Ocean

A Canadian Digital Twin of the Ocean would be a shared digital platform that brings together all of Canada’s ocean data, science, and models into a single, accessible digital environment.

Canada already generates significant ocean science and data through Fisheries and Oceans Canada, including various climate modelling, observation, scientific analysis and socio-economic analysis. However, this information remains fragmented across systems and is not easily accessible nor interoperable.

The goal is to create a foundation that enables users to build and apply tools for decision-making, analysis, and scenario testing.

The Digital Twin of the Ocean would act as a foundational platform, integrating existing datasets with shared computing and modelling capabilities. This platform would not be a single-use tool but an enabling environment for decision-making, analysis, and scenario testing, where “layers” can be built to support different applications and understand of the impact of decisions on both the marine environment and socio-economic activities.

Submitted by Nicole Uloth at Fisheries and Oceans Canada

AI Smart Forms & Data Reuse Platform

Government services require citizens and businesses to complete numerous forms, across many different departments in the Government of Canada, to access services.  

However, as forms are designed independently, they often repeatedly request the same information – name, address, income, employment history – even though much of this data already exists within government systems.

This leads to long and complex applications, higher error rates, increased processing time, and unnecessary administrative burden for both citizens and public servants.

The proposed solution would solve this problem by creating an AI-powered platform that identifies redundant form fields and enables secure reuse of existing government data to create simplified, pre-filled “smart forms.”

These would use AI to map and compare form fields across departments, and enable secure pre-filling of forms using existing government data with citizen consent. This would create a scalable platform that can be adopted across multiple departments with responsible, ethical, and secure use of AI.

Submitted by Sakshi Gupta at Canada Revenue Agency

AI Agent for Exploring Government Data

Many government websites contain valuable information and untapped insights, but these are hard to access or locate in government websites, creating a barrier for both public servants and the general public.

This project would create an AI-powered agent that helps users easily find and understand public data from Government of Canada sources such as Open Canada and Statistics Canada.

Users would be able to ask questions in natural language, and the agent would respond with clear answers based on the retrieved information, retrieving relevant datasets, reports, and files in real time.

The agent would provide more reliable answers than general AI tools such as ChatGPT by grounding responses directly in trusted government data, reducing AI hallucinations. It would also be able to reference and cite the exact datasets or reports used in each response, increasing transparency and trust, and making public data easier to access, understand, and use.

Submitted by Muadh Taghdi at Canada Revenue Agency