Winnipeg, MB – The City of Winnipeg will be moving forward with six innovative pilot projects to improve customer service through the use of artificial intelligence (AI).
These projects include:
- An automated chat agent for 311 inquiries
- An automated voice agent for Winnipeg Police Service non-emergency calls
- An automated development plan evaluation tool for permits
- Road condition monitoring through cameras mounted on City vehicles
- Voice sentiment analysis
- Automation of invoice processing
“At the State of the City Address in February, I issued a challenge to our Public Service to deliver a pitch by July 1 to use AI to improve customer service, and they delivered in a big way,” said Mayor Scott Gillingham. “These pilot projects have the potential to enhance service delivery and help staff do their jobs more efficiently, so I’m excited to see them launch in the coming months.”
Over the coming months, details on each of these pilot projects will be finalized including launch dates. Each pilot project will proceed through planning, procurement, analysis, implementation, deployment, and closing phases. A project may be changed or cancelled at any of these phases as the City learns about the benefits, costs, risks, and implications of a particular solution and adjusts.
Backgrounder
The following are outlines of the pilot project concepts. These projects may pivot during implementation to produce other benefits or mitigate risks as they are encountered.
Chat Agent
The City of Winnipeg currently provides information and fulfills service requests in a number of places, including in-person, through phone, and online.
The automated Chat Agent will add an additional way to access information and submit service requests through a multilingual chat agent that can be accessed through the City’s website.
The pilot implementation of a chat agent provides a conversational source of information, with knowledge of all content published on the winnipeg.ca website. The chat agent will communicate with customers in the language of their choice and is always available.
Our goal is to create easier access to information about City services, along with faster responses to some customer inquiries. It should also provide improved communication options for residents who do not speak English or French as their first language.
Voice Agent
The City currently operates multiple voice-oriented contact centers including 311, two 911 contact centers, a Winnipeg Transit Plus contact center, and a Water & Waste billing contact center. Each contact center strives to provide high quality service while minimizing wait times and optimizing costs.
This pilot project will initially be launched on the Winnipeg Police Service non-emergency phone line, and will divert some of the calls from a live human agent to a voice agent enabled by generative AI.
Modern AI-enabled voice agents can provide comprehensive and concise information, fulfill transactional service requests, and gather background information prior to the customer speaking with a live agent. They are always available, there is never a wait, and for some problems they can provide initial service while a customer waits for a live agent.
Our goal is to free up 911 operators’ time and energy to focus on emergencies, divert call volume, minimize overtime, improve morale and engagement of 911 operators, and improve access to data for call analysis insights.
Permit Automation
The City approves over 4,000 new building permits annually, covering all types of construction, from single family dwellings to apartments buildings. Submitted building designs are required to fall within the parameters defined in the various zoning by-laws and building codes. Often, the process involves back-and-forth communication between the applicant and the permit technicians reviewing their submission.
This pilot project of an automated development plan evaluation tool will pre-check designs and compliance with all relevant zoning regulations (i.e. setback, heights, frontage standards, floor area, etc.). An immediate report will be generated highlighting the passes and fails in the submission.
Our goal is to speed up the evaluation of some permit applications, accelerating the permitting process, and saving time and resources for developers and City staff.
Asset Condition Monitoring
The City delivers a broad range of services using physical assets located in the public right-of-way. Each of these assets must continually be maintained, serviced, and repaired.
The City also operates many cameras mounted to buses, bylaw enforcement vehicles, and other vehicles that are continually travelling city streets in predictable patterns.
This pilot project will identify and score potholes using cameras mounted to Winnipeg Transit buses on regional roads, residential, and connector streets. Advanced machine-learning enabled software will be used to analyze the imagery and produce actionable insights for management and maintenance teams.
Our goal is to have faster resolution of maintenance incidents like potholes, better communications to customers of previously reported incidents, and more informed decisions on right-of-way management.
Voice Sentiment Analytics
Voice calls to the City contact centers produce a large volume of audio data for analysis and learning. Audio data can be used to enable analysis and analytics with recommendations to improve contact center performance.
One area of opportunity for innovation is using this data to improve the well-being of emergency call-takers in our 911 call centres, who respond to repetitive traumatic situations and are at high risk of mental health injuries.
This pilot project will use an analytical AI agent to monitor emergency calls made to the Winnipeg Fire Paramedic Service’s 911 contact center to identify when a call taker may benefit from well-being intervention.
An AI agent can potentially identify changes in language, volume, and response time, enabling actionable insight for supervisors. An AI agent is not subject to mental health injuries, and this will further reduce harm to our staff by avoiding humans doing this work and experiencing vicarious trauma.
The pilot project will analyze completed or simulated calls, but the solution could also be used in real-time or near-real-time.
Our goal is to improve employee well-being, avoid mental-health injuries, and lower turnover rates.
Invoice Processing Automation
The City manually processes over 10,000 invoices per month. Manual invoice processing is time-consuming and can be prone to human error. Volumes fluctuate over the annual cycle, and there are times when the City is at risk of missing payment due dates.
This pilot project will apply machine vision and processing automation to accurately classify the elements of an invoice, populate financial systems, and reduce manual review and data entry.
Our goal is to deliver faster processing of invoices and more efficient operations.
“We are always looking for ways to positively improve the delivery of City services and are looking forward to launch these pilot projects,” said Tyler Gooch, Director of Innovation and Technology. “We know AI isn’t perfect, and we will be closely monitoring these projects to see if the benefits continue to outweigh the risks, and if the pilot projects meet the intended goals.”
The cost of each pilot project will not exceed $75,000 and will use existing contracts where available.
Once the pilot projects wrap-up, fully costed business cases will be forwarded to the annual budget process.