Module
The spec says FD60. The model says 30.
Bimiq reads your specification clause by clause, turns every checkable requirement into a real test, runs it against the live Revit model, and shows you exactly which elements disagree, with a fix that won't break the ones that were already right.
Your spec is prose. Your model is a database.
A specification is never checked directly. It's paragraphs of intent, written for a human. Bimiq turns each clause into something testable: a scope that selects the elements, and a requirement they have to meet.
The clause, as written
08 11 00 (2.3.B)
“Fire doors serving stair cores shall be rated FD60 and shall be self-closing.”
What Bimiq extracts
Doors · doors serving stair cores
Fire Rating ≥ 60
type · writablecriticalconfidence 92%
Doors · doors serving stair cores
Self-Closing = Yes
type · writablecriticalconfidence 88%
One rule per distinct requirement. A clause carrying two obligations becomes two tests, because that's how many ways it can fail.
Four hundred pages. Every clause.
The entire document goes through, in overlapping passes so a requirement split across a boundary isn't lost. Duplicates collapse: the same rule written twice, or the same spec uploaded twice, is still one rule.
The document
Overlapping windows: nothing falls through a boundary, nothing is counted twice.
It doesn't summarise, and it doesn't stop early. You get every distinct requirement, not a précis of the ones that looked important.
What it won't claim
The honest half of the answer.
Some requirements aren't model data.
“Two coats of primer.” “Submit certificates.” Site testing, workmanship, procurement. There's no parameter for any of it, and Bimiq doesn't invent one. These land in Manual review: listed, tracked, never silently dropped.
When the mapping is uncertain, it steps back.
If a requirement can't be bound to a real category and a real parameter with a concrete value, Bimiq doesn't guess. It moves the rule to manual review and tells you why: couldn't be bound to a specific model parameter.
Confidence is on the label.
Every mapping carries a score. Anything below 60% is flagged for you to confirm before it counts against your model. You're reviewing Bimiq's reading of the spec, not taking it on faith.
Then it asks the model.
Bimiq queries the live Revit model for the elements that break each rule. One requirement: invert it, and the violators come back directly. Several: subtract the fully compliant from the scope.
Either way the number is exact, not a sample or an estimate. Every offending element comes back with its current value.
Runs against the open model through the Bridge. Revit 2020-2027.
Doors · Clause 08 11 00 (2.3.B)
CriticalFire doors serving stair cores shall be rated FD60.
14 of 47 doors non-compliant: 30 on 14 doors
Zero matching elements is not a pass.
If the model has no stair-core doors at all, Bimiq reports the rule as not applicable, and it gets its own tab. Calling an empty result “compliant” would be a lie, and it's the lie that gets found on site.
The fix
Fire Rating lives on the type.
Set it on your 14 violating doors and you've just changed every door sharing those types, including the 33 that were already correct. That's a fix that passes the check and fails the project.
So Bimiq offers both, and makes you choose.
Set the parameter
Writes the value straight onto the violating elements. Fast, and correct when the parameter is per-instance.
Duplicate the type
Creates a compliant variant of each affected type and moves only the violating doors onto it. The 33 that were already right never move.
It re-checks after writing.
“Fixed 12; 3 still non-compliant.” No fix is assumed to have worked.
Every write is logged.
What changed, on how many elements, by whom.
When it can't be written, it says so.
Calculated values like glazing U-value or computed areas aren't writable, and Bimiq won't fake it. Raise an RFI to confirm, with the clause and the model's actual values already attached.
Your next coordination cycle starts soon
Connect your model, drop in your specs, and see what Bimiq finds in the first hour.
14-day free trial. No credit card required.