Technology & Methodology
Deterministic First.
AI Second.
Most AI compliance tools run natural language processing across code documents and return probabilistic answers. BIM Pro AI does not. The platform separates two fundamentally different problems.
Problem 1
What does the code require, and does this building meet it?
Deterministic compliance engine
Not AI
Problem 2
How do we explain that result in context?
Citation-aware AI layer
Scoped and constrained
This separation is not a technical detail. It is the product architecture.
Code Document Processing
User-Supplied Energy Code Documents
The AI knowledge layer is designed to process energy code documents that users supply from their own authorized sources — not a static internal copy maintained by BIM Pro AI.
When a user provides their applicable code document, the system processes it in-session to retrieve specific provisions by section reference and generate citation-aware explanations tied to computed results.
Uploaded code documents are not stored, indexed, or retained after the session ends. See our Standards Access and Content Policy for detail.
Citation Architecture
Traceable to Section
AI Guardrails
Constrained by Design
Our Approach to AI
AI in a Carefully Scoped Role
We use AI to generate explanations and retrieve relevant code citations based on calculated results.
We do not present probabilistic AI outputs as authoritative compliance conclusions. This approach is intentional. In professional compliance workflows, auditability is not optional.
AI Does Not
Example Output
What This Looks Like in Practice
A BIM Pro AI output for a commercial building envelope compliance check includes:
- —The calculated U-value for each envelope assembly, derived from Revit model geometry
- —The prescriptive envelope requirement for that assembly type in the project's climate zone, cited to the governing provisions in your supplied code document or adopted public code
- —A pass/fail determination produced by the compliance engine
- —A plain-language explanation of the result, citing the specific section, generated by the AI layer
What it does not include:
- —Any value the deterministic engine did not calculate
- —Any claim the AI layer extended beyond what was computed
- —Any output that cannot be traced to a specific model input or cited provision
Questions About Our Methodology?
We're happy to talk through the technical architecture with engineers, reviewers, or potential partners.