National & Ontario building code
Approvals, without the guesswork.
Civixa reads your model or permit set, then judges it against the code with a deterministic engine — and cites the clause for every answer it gives.
- Reads
- Your model, or your drawings
- Cites
- The clause, every time
- Never
- Guesses
02Why it is different
The AI reads. It never decides.
Compliance is not a judgement call a language model should be making, and everyone shipping “AI code review” has quietly decided otherwise. Civixa is built the other way round — and that single decision is why its answers hold up.
Everyone else
- Asks a language model whether something complies
- Gives you a confident answer either way
- Changes its mind between runs
- Leaves you checking its work
Civixa
- The model reads. It never gets a vote on compliance.
- Judgement is arithmetic against the written code
- The same input gives the same answer, always
- Every result carries the clause behind it
An answer you have to double-check is not an answer. It is homework.
03No bluffing
It would rather say nothing than say something wrong.
Every other tool in this category has the same failure: when it isn’t sure, it answers anyway. You find out at the counter.
Civixa holds. It marks what it could not stand behind, tells you what it needs, and gives you the answer the moment it has it — so a pass from Civixa means something.
“I’m not certain enough to call this.”
04The reckoning
Nothing slips through the gaps.
Most tools hand you a list of problems, which quietly merges “this is fine” with “nobody looked at this.” Civixa refuses to blur the two. Every element comes back with one of four answers — and silence is not one of them.
It clears the requirement
Measured, compared and cleared — alongside the clause it was measured against.
It falls short
You get the requirement and the shortfall, before a reviewer finds it.
Something underneath is unresolved
Never quietly passed. It waits, and names exactly what it is waiting on.
No rule covers it
Said out loud, so an unchecked element never sits there looking compliant.
The difference between a tool you check behind and a tool you can rely on is whether it admits the gaps.
05What you get back
An answer that survives the question “why?”
Anyone can flag a door. The hard part is standing behind it when a reviewer, a client or your own principal asks where that came from — and that is the part Civixa is built for.
The clause, not a hunch
Every answer arrives attached to the sentence of the code it came from — the article, and the requirement it sets.
The number it measured
Not “this looks wrong”. The dimension it read, beside the dimension the code asks for.
The same answer twice
Run it again next month and the verdict is identical. Nothing drifts between one run and the next.
Something you can hand over
A report a reviewer can follow, a colleague can check, and you can defend in a meeting.
06Two ways in
The model, or the drawings.
IFC · straight from your authoring tool
It reads the building, not a spreadsheet
Export from Revit, ArchiCAD or anything that speaks IFC. Civixa works from the model's own geometry and spatial structure — the rooms, the storeys, the openings — so it is measuring the building you actually drew.
- Works from
- The model's own geometry
- Understands
- Storeys, rooms, openings
- No prep
- No tagging, no cleanup pass
07Built for here
Canadian codes, taken seriously.
Written for the codes you actually submit against
The National Building Code and the Ontario Building Code, in the language and the units your set is drawn in.
It grows without waiting for us
Requirements are data, not code. New parts and provincial amendments are added without rebuilding the engine.
Made for the way permits really move
Built with the people who submit sets, get comments back, and resubmit — not as a demo of what AI can do.
See it on your own set.
Bring a model or a permit set and watch it come back with the clause attached.