California recently created laws that mandate municipalities to deliver rapid permit approvals for rooftop solar projects.Damian Dovarganes/The Associated Press
Vancouver modernist architect Arno Matis was grappling with a problem that is table stakes for his profession, but has, in recent years, been ratcheted up to the level of national crisis by politicians pressing for solutions to the housing affordability crisis: the amount of time it takes to secure a building permit, which seems to grow longer by the year.
“What we were finding was that as regulations were getting more and more complex, and different departments were getting created in various municipalities, the time to get through all of this was getting multiplied quite dramatically, to the point where, in some cases, we were seeing the approval timeline was longer than it actually took to build the building,” says Mr. Matis, who set up his own practice 20 years ago after working for Bing Thom.
A mid-size project in Vancouver could be bogged down in the approvals process for two or three years. “We saw that there was an issue there,” he continues. “Obviously, our clients were getting dragged along for the ride.”
Major employers in tech and finance expect new hires to be ‘AI literate’
While Conservative Leader Pierre Poilievre blames municipal “gatekeepers,” Mr. Matis understands that those delays result from an accumulation of zoning policies and technical building and energy code requirements, all of which bog down the municipal officials tasked with vetting projects. Casting around for ways to improve the process, he reckoned that specialized digital automation tools, including artificial intelligence, might speed the proverbial plow.
To that end, Mr. Matis’s firm last fall launched a technology-development partnership with an Australian firm, Archistar, that has built AI tools capable of checking whether project proposals comply with zoning rules, building codes and other planning policies. Founded in 2010 by architect Benjamin Coorey, Archistar now provides its software to municipalities, developers and architects in California, Texas, New York, B.C. and Edmonton.
The company is by no means alone in this rapidly emerging sector. Other tech firms such as San Francisco-based Symbium, Spokane-based GovStream.AI and Chicago’s CivCheck have rolled out AI-based platforms for automating permit approvals. California is regarded as the leader because of recent state laws that mandate municipalities to deliver rapid approvals for rooftop solar projects and accessory dwelling units (known in Toronto as garden suites).
Workers assemble an accessory dwelling unit for Eaton Fire survivors in Altadena, Calif., in December.Mario Tama/Getty Images
The City of Toronto is in the midst of vetting software firms that can provide a service to partly automate the prescreening of development applications, Kamal Gogna, the chief building official, said in an interview earlier this month. A pilot version will launch in the fall.
The goal, say those who’ve worked with these technologies, is to allow planners to spend more of their time on the parts of the approvals process that require their professional judgment instead of slogging through highly repetitive tasks associated with routine permits.
“Ultimately, the idea is to use the technology tools to reduce the bottlenecks in this process,” Mr. Matis says. “The way municipalities process permits today is very much the way they’ve been doing it for 100 years.”
While both architects and planners have long used all sorts of advanced software tools, such as AutoCAD and geographic information systems (GIS), the tasks involved in determining whether a development application conforms to zoning, building code and planning rules have stubbornly resisted automation. It’s not a matter of simply uploading plans into ChatGPT or some other AI and then asking if they comply with local regulations.
AI adoption in commercial real estate still in early stages, expert says
The reasons are highly technical, but have to do with the archaic nature of both plans and bylaws. While they can be digitized, blueprints or working drawings are still two-dimensional diagrams, while municipal regulations exist only in written format, not as a set of mathematical formulas translated into computer code.
AI-based permitting software is designed to figure out, for example, whether a proposed garage is too close to the property line or if a garden suite satisfies the zoning and setback rules that apply at the address where the project is proposed. To do that, the software must be able to read and interpret the documents that planners rely on: bylaws and architects’ drawings. Mr. Matis’s AI team and Archistar are building software tools that can bridge that digital gap even before plans are submitted to the municipality, so designers and their clients can make the necessary adjustments ahead of time in order to reduce time-consuming back-and-forths.
Some California municipalities have set up online application portals for permits for standardized projects, like rooftop solar installations and ADUs. Various AI-permit software systems working in the background flag details that are non-compliant. These systems can cut the permit processing time from weeks to minutes, depending on the complexity of the project.
The City of Kelowna, considered an early adopter of AI, allows homeowners, contractors and builders to apply for permits through such portals, with the AI performing tasks such as prompting users to add missing data or verifying technical details, like setbacks. But, according to Kelowna’s chief technology officer, James McGregor, the system isn’t yet set up to automatically issue a building permit, as happens in some California cities. “There’s [a] human in the loop,” he says. “We take the approach with AI that it’s not about replacing people.”
A gold-standard garden suite, dressed in copper, in Toronto’s Bridle Path
Toronto is taking a somewhat different tack, envisioning a voluntary AI-based system that prescreens any sort of development application for zoning compliance. “Individuals who go through using this proposed technology will be able to receive quicker responses to their initial submission,” says Mr. Gogna, who estimates the system will be able to provide an answer within hours, compared with the 10- to 30-day period that is currently typical. The rest of Toronto’s application review, however, will still be completed by planners and code examiners.
Pamela Robinson, a planning scholar at Toronto Metropolitan University who studies urban technology, stresses that such software tools shouldn’t be seen as some kind of panacea for excess bureaucracy. Rather, she says, they’re about boosting productivity within planning departments, so professional planners spend less of their time on routine tasks in a period when city-building has become increasingly complex.
“If you can reduce the time people spend checking boxes, what else could they work on?” Prof. Robinson notes. “We all need to get better at making it easier for applicants to both get into the process, but also for the process to move along at the speed of common sense, [and] in a way that’s not dehumanizing.”