When a new edition of a building code is published, engineers must understand every change that affects their practice. Traditionally this means reading both editions side by side - a process that takes days for a major code like the NBCC or BCBC. I built a tool that automates this comparison at the individual clause level, with AI-powered summarization that tells you not just what changed, but whether it matters.

My Role

Sole developer. Self-initiated project - identified the need during preparation for NBCC 2025 code changes and built the solution independently.

What I Built

Point-by-point diff engine:

  • Parses building codes into individual clauses and sub-clauses
  • Compares each point between the old and new editions, identifying additions, deletions, and modifications
  • Handles structural differences in numbering and organization between code editions

AI-powered change summarization:

  • Each identified change is summarized by an AI model that explains what changed and why it matters for engineering practice
  • Summaries are written in plain language, focusing on practical impact rather than legal text differences

Editorial vs. substantive classification:

  • Every change is automatically classified as “editorial” (rewording, renumbering, formatting) or “substantive” (changes that affect design requirements, loads, factors, or procedures)
  • Engineers can filter to see only substantive changes, saving significant review time on codes where the majority of changes are editorial

Impact

  • Reduces code comparison from days of manual effort to automated analysis
  • Used for the NBCC 2025 vs. NBCC 2020 comparison - the results formed the basis of a technical presentation delivered to all structural engineers in the company
  • Applicable across building, electrical, mechanical, and structural codes - any standard published in a parseable format
  • Enables engineers to focus their review time on changes that actually affect their design practice