An early build. It has not been run against a production model by anyone yet, so point it at a job you already know rather than one you cannot afford to have opinions about.
A review aid, not a compliance certificate. It cannot approve a design, and a report with no failures is not evidence of compliance.
A Revit add-in that measures your model against thresholds held in editable rule packs, then shows you every place it falls short.
Three answers, not two. Fail, Pass, and Inconclusive when the model cannot answer the question at all. A run full of Inconclusive is telling you the model is not yet modelled to a standard anyone can check, which is worth knowing early.
Pick your rule packs and your state, run the check, and read the findings grouped by rule. Click any one to jump to that element, or paint the whole set into the active view.
Tick what you want fixed. Parameter changes are applied as one undo step, with the count of other elements each will move shown first. Anything needing a wall moved is walked through one at a time. Then it re-checks and tells you what actually closed.
An early build, changing quickly. Only what you would notice is listed.
Findings are grouped by the change that closes them. A real building produces hundreds of items, and they are rarely hundreds of problems: sixty doors failing a width are one door type and one write. The list now leads with the change — “widen door type to 1 125 mm, closes 60” — and opens to the findings under it. On a four-storey office we test against, 820 items to confirm read as nineteen lines, and 90 failures as six.
Fix, set aside, zoom and colour on every row. And on every group, so a change that closes sixty findings is one button rather than sixty. Closing a single finding used to mean ticking it, going up to the toolbar and clicking; it is one press now.
Three fixes worth naming. Comparing this run against an earlier one failed whenever something had been fixed — which is the run you would want to compare. A rule that says several things about one building, like the sanitary counts, kept them under one record, so setting aside the closets set aside the urinals. And the panel that finds shared changes was matching on the element rather than the type it belongs to, so on a real model it found two changes where there were forty.
Send items to a consultant as a page they can read. Handing fire services or energy to whoever owns it now produces a document that opens in any browser: the requirements listed plainly, a box against each one, and a button that writes the reply for you to import. Their answers close the items in their name.
All six of the Board’s calculators, built in. Sanitary facilities, lighting power, glazing shading, wall R-values and whole-of-home join the gutter calculator — six spreadsheets you would otherwise open and retype after every revision.
Sanitary facilities is the one to look at first. It works out how many closet pans, urinals and basins Part F4D4 asks for at your occupancy and puts that beside how many are drawn.
Glazing shading shows you the problem early. Every window gets the multiplier for your climate zone and the elevation it sits on, so the west-facing glass driving your result is obvious well before an assessor tells you.
It says “check”, not “audit”. An audit is something a registered person signs. This is not that.
Roof drainage, against the Board’s gutter table. Each roof’s catchment is measured and looked up using the design rainfall for your state. Above 70 m² it tells you the Deemed-to-Satisfy table does not reach that far and the sizing needs a hydraulic design.
Fewer findings that were never really wrong. A room modelled at exactly 2400 mm could measure 2399.9997 and be reported as a failure. It no longer is.
Wind region, from the project location. The AS/NZS 1170.2 region is read from Geoscience Australia’s published map, so the tool knows whether your site is cyclonic or alpine instead of raising every wind and cyclone provision on every project. Twenty-four rules now scope themselves.
It tells you when the Code has moved past it. These packs are NCC 2022. If your project is somewhere NCC 2025 is already in force, or in a transition where either applies, it says so at the top of the results rather than quietly measuring against the wrong edition.
Climate zone from the published map, read from the project location rather than estimated from the nearest capital city — which was wrong for a good deal of regional Australia.
Revit 2024, 2025 and 2026. Two builds; pick the one that matches your Revit.
Hand work to the consultant who owns it. Most of what a model cannot settle is not the architect’s to answer. Export an area, send it to whoever did that package, and import their reply.
It reads where the building is. Setup fills in the state and climate zone from the project location. Several provisions apply only in certain zones, and before this they were raised everywhere or nowhere.
Compare with an earlier run. What was already failing before a revision is not what the revision changed. Measure this check against the one you ran for DA and see what broke, what was fixed, and what moved.
First build. Revit 2026 only.
To remove it, delete those two items from the same folder. Nothing else is installed and nothing is written outside your user profile.
Take the download that matches your Revit version. There is one for 2024 and one for 2025 and 2026; they are not interchangeable, and the wrong one fails at start-up with an error about a missing .NET file rather than anything about the version.
This is the first build and nobody has used it on a real project except me, so anything you find is useful - including “it would not install”. Work down in order and stop wherever it falls over; there is no point testing step five if step two failed.
You are not expected to write anything up. A sentence per step, or a screenshot of whatever looked wrong, is plenty.
Follow the five steps above. The one that trips people up is the unsigned-add-in warning on start-up - that is expected, and Always Load is the right answer.
Working if: Revit opens normally and there is an NCC tab on the ribbon, with an Check Model button on it.
Open a real project - ideally something with a few levels, some doors and a stair, not an empty template. Click Check Model.
The first time, it asks three questions: what kind of building, which state, and which areas somebody else is looking after. Answer them for the project you have open. The count at the bottom of that dialog should move as you change answers.
Working if: A window opens with a list of findings in it, and Revit is still usable behind it.
Have a poke around before doing anything careful. Type something into the search box. Switch Show between Fail, Inconclusive and Pass. Change Group by. Open Settings and Help and close them again. Drag the window, resize it, minimise it, close it and press Check Model again.
Working if: Nothing freezes or throws an error, and reopening brings your check straight back rather than running the whole thing again.
This is the part only you can judge. Read down twenty or so findings and ask, as the architect: is that actually true of my model?
Most findings will say Inconclusive. That is deliberate - most of the Code is about materials, ratings and certification that a model does not carry, so it raises them for you rather than staying quiet. Tell me if that volume is useful or just noise.
Working if: The numbers it reports match what you would measure yourself.
Do this on a copy of a project, not live work. It writes to your model.
Tick a finding that has a spanner on it and press Fix. It should change the parameter and the finding should turn green. Then press Bring back to undo it, and check in Revit that the parameter really went back. Ctrl+Z should also work normally.
Some findings need geometry moved instead. Those tell you the change required - needs geometry moved - but do not touch the model themselves. That is on purpose.
Working if: The parameter genuinely changes in Revit, and undoing it genuinely puts it back.
Go along the row of buttons and try each one:
Working if: Every button does the obvious thing, and Clear leaves your view exactly as it was.
Export writes a report you can send on. Read a sample report before you install anything - it is the real document, generated by the demo.
Use a copy of a project for step five. The fixes write real parameter values into the model. Everything it changes is recorded and can be put back, but there is no reason to prove that on live work.
It is not a certifier and the thresholds have not been checked by one. If it disagrees with you about the Code, assume you are right and tell me - that is exactly the thing I need to hear.
Send it back however you normally reach me. If it crashed, the useful thing is what you were clicking when it went.