Ten small jobs and one decision
Take a DN65 pipe through a beam, the most ordinary clash there is. Done by hand, it breaks into ten jobs, and only one of them needs a coordinator's experience:
- Find it in the report: the row, the test, the two element ids.
- Locate it in Revit: open a view, section box, find the right link.
- Judge which service moves and how. This is the decision.
- Check the space around it: above, below, either side, the ceiling, the next service.
- Split the run either side of the obstacle.
- Offset the segment by the distance you measured.
- Add the fittings, 45° where there is room, 90° where there is not.
- Reconnect and check every joint.
- Check links and new clashes, because the fix itself can clash.
- Undo and retry when it does.
The split above is illustrative, but the shape is what matters: judgement is one slice, and the rest is work that has to be done exactly right and done again whenever it goes wrong.
The costs that do not show on the clock
The fifteen minutes are only the visible part. The expensive part of a clash is usually what happens after the fix:
- The fix that clashes. A pipe dropped under a beam lands on a cable tray two metres along. Nobody sees it until the next test run, a week later.
- The wrong mover. A supply duct rerouted around a small pipe looks tidy and then needs the HVAC engineer to re-check the airflow. The pipe should have moved.
- The forgotten re-check. A sprinkler branch with a new low point and no auxiliary drain; a heating pipe with a new high point and no air vent.
- The same clash, five times. One badly placed main appears as five rows; each is fixed separately, and the fixes fight each other.
- The context switch. Each clash means a new view, a new section box, a new link to find. A coordinator working a list of 300 does that 300 times.
Each of these is cheap to avoid at the moment of the fix and expensive later. That is why a tool that only automates the drawing saves less than it seems: the routine it skips was never the whole cost.
The arithmetic, with your own numbers
The calculator on our home page uses three lines of arithmetic, so you can check it:
hours back per month = clashes per month × (minutes by hand − minutes with ClashFree) ÷ 60
working days a year = hours back per month × 12 ÷ 8
money per month = hours back per month × hourly cost of a coordinator
With 15 minutes by hand, 3 with ClashFree and a coordinator at $60 an hour, all three illustrative:
| Team | Clashes / month | Hours back / month | Days / year | Per month |
|---|---|---|---|---|
| One coordinator, one project | 120 | 24 | 36 | $1,440 |
| A small MEP team | 300 | 60 | 90 | $3,600 |
| Several projects in coordination | 1,000 | 200 | 300 | $12,000 |
Two honest caveats. Three minutes per clash is the beta's design target, not a measured figure, and some clashes will always take longer because they need a person: flags, other trades' calls, an engineer's re-check. On the other side, the table counts none of the hidden costs above, which is where most of the real savings tend to be.
A week of clash reports, two ways
Picture a Monday. Two reports arrive from the weekly coordination run, Structure against MEP and Architecture against MEP, with 600 lines between them for your floor. By hand, the first hour goes to de-duplicating and sorting in a spreadsheet, and most of the week to working the list from top to bottom. By Thursday a third of it is done, and the fixes you made on Tuesday show up in Friday's run as new clashes with the cable tray.
With ClashFree, the import does the sorting: duplicates merged, no-action clashes set aside with their reasons, groups by culprit run, your trade's groups in To fix. You tick the first twenty groups and send them to AI, and while the queue works you review the ones it has verified: Keep, Reject or ask for a change. The fixes are checked against the host model and every loaded link before you see them, so Friday's run has fewer surprises. The groups that need another trade sit in Coordination, with the reason, ready for the meeting.
The time saved is real, but the bigger change is the order of work. You spend the week on decisions instead of on the routine around them, and the routine is done the same careful way on the five-hundredth clash as on the first.
What software should take
Everything that has a right answer and has to be done exactly: locating the clash and both runs, measuring the free space with the clearance of the clash test, listing every feasible fix, splitting, offsetting, fitting and reconnecting, checking for new clashes against the host and every link, and rolling back cleanly when a check fails. Also the arithmetic of judgement: what each change costs on that system, which fixes break a hard limit, which run is the culprit of five clashes. That is what ClashFree does.
What it should leave with you
The decision to keep a fix. The engineering consequences, which ClashFree lists with every fix (airflow, hydraulics, falls, firestops) and never claims are fine. Another trade's runs, which it never edits. And the cases where no fix passes, which it flags with its reasons instead of forcing an answer. A tool that takes those away from the coordinator is not saving time; it is moving risk.
Imports the report, groups it by cause, and for every clash you send to AI lists and scores the options, has the AI choose and explain, builds the fix, verifies it in Revit and rolls it back. A queue does this ahead of you, so your three minutes are reading, previewing and pressing Keep.
How the beta will measure it
During the beta, the add-in's usage analytics record per clash, without names or model content, how long it takes from the first proposal to your decision, how often the first proposal is kept and how many revisions a kept fix needed. When there is enough of it, we will publish the measured numbers on the home page in place of the design target. If you join the beta, your clashes are part of that measurement.
Next: see the loop that produces those three minutes in AI-native clash coordination, or measure it yourself in the free beta.