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 FAQ for detail.

Citation Architecture

Traceable to Section

Every AI-generated explanation cites the specific section reference that supports it. The system tracks citations, not confidence scores — so every explanation is verifiable against the source document rather than inferred from pattern matching.

AI Guardrails

Constrained by Design

Guardrails prevent the AI from making claims beyond what the calculation layer determined. The system is designed to explain — not extend or extrapolate. Specifically: if the compliance engine has not produced a result for a given parameter, the AI layer will not generate an explanation for it. Scope is enforced architecturally, not by prompt engineering alone.

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

Perform compliance calculations
Make code determinations
Produce outputs independent of the deterministic engine
Generate unsupported claims or extrapolations

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.