The building permit should be a CI/CD pipeline.
Who This Matters To (And Why)
Critical: Developer,City,Architect. These parties make or lose money directly based on this thesis.
Important: GC,Banker,Inspector. These parties execute decisions shaped by this thesis.
Context: Investor,Engineer,Broker. These parties need to understand it to avoid friction.
Highest typology impact: Multifamily,Office,Mixed Use,Urban Planning. Lower impact: Industrial,Retail.
The building permit should be a CI/CD pipeline. Submit, review, approve, deploy in days not years.
How It Shapes Development
The building permit should be a CI/CD pipeline because it is a compliance check that currently runs once, at the end of the design process, when errors are maximally expensive to fix. CI/CD — continuous integration and continuous delivery — is the software practice of running automated tests on every change, not just at release. A building permit process structured as CI/CD would run compliance checks continuously throughout design: every time a wall moves, every time a floor plate changes, every time a unit is added. Non-compliance surfaces immediately, when the design team can fix it cheaply, rather than at permit submission, when it requires redesign.
Current permit review is batch processing. The design team submits a complete set of documents. The jurisdiction queues the review. After weeks or months, comments come back. The team responds. Another review cycle begins. Each cycle takes weeks. A project with three review cycles has spent 3–6 months in permit processing. A CI/CD permit process would eliminate the batch cycle: compliance checks run continuously, comments surface within hours of each design change, and the permit issues when the compliance check passes cleanly. The queue disappears because there is no batch submission.
Machine-readable zoning codes make CI/CD permitting possible. If the zoning code is a structured dataset — setbacks, height limits, FAR, use restrictions, parking ratios expressed as queryable parameters — then a building information model can be checked against the code automatically. The model knows its site boundary, its building footprint, its floor areas, and its use categories. The compliance engine queries the zoning dataset and returns pass/fail for each requirement. This is not theoretical: jurisdictions that have digitized their zoning codes can already do this for simple cases. The limitation is that most zoning codes are PDFs, not structured data.
Permit as a service is the endpoint. A developer in a jurisdiction with a CI/CD permit process uploads a building model, selects the applicable code version, and receives a compliance report within minutes. When the report is clean, the permit issues automatically. The human reviewer is reserved for cases that require judgment: variances, discretionary approvals, novel construction types. Routine compliance verification is automated. This is not a displacement of the regulatory function — it is an acceleration of the routine portion of it, freeing reviewers to spend their time on cases that actually require human judgment. The bottleneck in current permit review is not reviewer capacity on hard cases; it is reviewer time spent on easy cases that could be automated.