Drafting and reviewing urban planning bylaws is among the most time-consuming work in Quebec municipal departments. A zoning bylaw can run to several hundred sections, accumulated through successive amendments over several decades. The internal consistency of these documents weakens over time, cross-references multiply, and revisions turn into full-scale projects that tie up professional resources for months.
Artificial intelligence does not solve this problem with a wave of a magic wand. But used methodically, it can substantially reduce the time spent on certain steps of the process, particularly summarizing, spotting inconsistencies and producing drafts. This page sets out what AI can concretely do in this context, what it should not do, and a working method you can apply today in a Quebec urban planning department.
If you want a general overview of AI applications in planning practice, the page AI and urban planners in Quebec: how to use it in your daily work is a good starting point.
What AI can really do with a zoning bylaw
It helps to separate reasonable, documented uses of AI from what belongs more to technological fantasy. In regulatory drafting, the concrete benefits are real but limited in scope.
Summarizing and condensing long documents
Feeding a complete zoning bylaw to an AI tool to obtain a summary structured by theme is one of the most reliable and immediately useful applications. The tool can extract the main zone categories, permitted uses by sector, key siting standards and special provisions, producing a readable summary document in a few minutes instead of several hours of reading.
This capability is especially valuable when onboarding a new team member, preparing a meeting with developers or elected officials, or quickly comparing two successive versions of the same bylaw to identify what changed from one revision to the next.
Spotting inconsistencies in an existing bylaw
Urban planning bylaws drafted over many years frequently contain contradictory provisions, references to repealed sections, or wording that no longer matches the current regulatory context. An AI tool can analyze the entire document and flag areas of potential tension: sections that contradict each other, terminology used inconsistently from one part to another, references to provisions that no longer exist.
The output is not a legal opinion. It is a preliminary screening that lets the planner focus on the problematic passages instead of rereading the entire document to find them.
First draft of new sections
When a new bylaw has to be written, or an existing one strengthened to cover a situation the current version does not address, AI can produce a first draft of sections based on a description of the objectives. That draft will never be ready to adopt as is, but it provides a structured working base that speeds up the drafting process.
To get useful results at this stage, the quality of the instructions given to the tool is decisive. That is covered in a separate section further down this page.
What AI should not do in this context
However capable the tools available in 2026 may be, certain parts of the regulatory process must stay entirely under human control. This is not excessive caution: it is a matter of professional responsibility and legal rigour.
Validating legal compliance
An urban planning bylaw in Quebec must comply with the Act respecting land use planning and development, with government planning orientations, with the objectives of the RCM land use plan, and with the provisions of the metropolitan land use and development plan where applicable. That validation requires an in-depth legal reading that AI cannot perform with the reliability an official regulatory act demands.
Language models can produce statements about a provision's compliance with the Act that are convincing in form but wrong in substance. A planner who signed off on a bylaw based on an AI-generated validation would be putting their professional liability on the line with no safety net.
Replacing discussions with elected officials and residents
Drafting an urban planning bylaw is a professional exercise, but it is also a political process. The planning choices it expresses come out of deliberations with elected officials, public consultations and trade-offs between interests that are sometimes opposed. AI can produce a technically coherent text without the slightest understanding of these local dynamics. The political and territorial content of a bylaw must therefore be defined by people before AI is brought in to shape the text.
Step-by-step method for using AI in regulatory drafting
The difference between a frustrating and a productive use of AI in this context often comes down to how you prepare the work before you even open the platform.
Preparing the right context in the instruction given to the tool
An AI tool produces results that match the quality of the information and instructions it is given. For a regulatory drafting task, the instruction has to specify several things: the type of bylaw involved, the municipality or type of territory concerned, the applicable legal framework, the planning objectives the bylaw must express, and the expected tone or register.
For example, an instruction like "draft a zoning bylaw section" will produce a generic output of little use. An instruction like "draft a zoning bylaw section for a rural Quebec municipality that wants to regulate the siting of accessory buildings on lots of less than 4,000 square metres in residential zone R1, providing for a maximum area of 20% of the area of the main building and a minimum setback of 1.5 metres from side property lines" will produce a result you can work with directly.
Reviewing and correcting the AI output
A draft produced by AI should be reviewed with the same level of attention as a text written by a capable intern who is unfamiliar with the Quebec municipal law specific to your territory. Wording can read smoothly while remaining legally imprecise. References to provisions of the Act or to provincial standards must be checked one by one. The terms used must be consistent with the vocabulary already established in the existing bylaw if the task is to add provisions to a document in force.
Does your team want to build this working method into its daily practice? The artificial intelligence training for Quebec municipal teams covers these techniques in depth with exercises drawn from real situations.
A concrete example: reviewing a subdivision bylaw with AI
The fictional municipality of Saint-Anselme-des-Laurentides has a subdivision bylaw adopted in 2008 and amended seven times since. Inconsistencies were flagged by a notary during a recent real estate transaction: two sections appear to set contradictory minimum lot area standards for lots in the resort zone.
The planner in charge uploads the full bylaw to Claude, whose context window can handle the document without truncating it, and asks it to identify every provision dealing with minimum lot areas, indicating where each appears in the document and flagging any apparent contradiction between the sections found.
In under two minutes, the tool produces a list of the relevant sections with their contents and does flag the contradiction between section 14.3 and section 22.1, which use different thresholds for the same thing depending on the part of the document they come from. It also suggests that three other sections may need terminology harmonization, without claiming these are legal inconsistencies.
The planner can now work on the two problematic sections in a targeted way, submit a proposed amendment to the municipal legal department and prepare a summary for elected officials, all from an initial screening that would otherwise have taken several hours of careful reading.
This kind of workflow also applies to permit application review processes. The page onAI for reviewing building permit applications presents a similar approach adapted to that other side of municipal planning work.
Frequently asked questions
Does a bylaw drafted with the help of AI have the same legal force?
Yes, provided it has been reviewed, validated and adopted through the usual legal processes. The legal force of a municipal bylaw comes from its adoption by the municipal council in accordance with the procedures set out in law, not from how the text was prepared. What matters is that the planner in charge exercised professional judgment on the content before submitting it for adoption.
Which document format works best with AI tools?
Native text files, Word documents and PDFs of digital documents give the best results. PDFs created by scanning paper documents often cause character recognition problems that degrade the quality of the analysis. If your bylaw exists only as a scan, converting it first with optical character recognition, then checking for the most common errors, will significantly improve the quality of the results.
Can AI produce bylaws in a format compatible with municipal management software?
General-purpose AI tools produce text. Final formatting in a document management system or a standardized regulatory template remains a manual step. Some municipalities have built Word templates into which AI outputs are dropped directly, which cuts down on layout work. To our knowledge, there is not yet any native integration between the major general-purpose AI platforms and the municipal regulatory management software used in Quebec.
Can AI be used to translate a bylaw into English for bilingual municipalities?
It is technically feasible and the results are generally solid for ordinary text. For an official regulatory document, an AI translation should be reviewed by someone with a command of legal English and a knowledge of Quebec municipal law in both languages. A translation that reads smoothly can still shift the meaning of technical terms that have no exact equivalent from one language to the other.
How long does it take for a team to become comfortable with these tools?
Field experience shows that with structured training and a few weeks of practice on real tasks, most planners reach a level of functional autonomy sufficient to fit AI into their usual workflow. Time savings become noticeable fairly quickly on large document-heavy tasks. The longest learning curve involves writing precise instructions, a skill that develops mainly through repeated practice.
