PROJECT 13 / AEC & MANUFACTURING
Draw an interior, walk through it, then build it
UX for ICE, the software that takes an engineering-accurate interior from 2D drafting through interactive 3D and VR to a manufacturing order.
- Role
- UX Designer
- Where
- DIRTT Environmental Solutions
- When
- 2018
- Links
- DIRTT
CONTEXT
The drawing is the order
ICE is not a visualisation tool that happens to be accurate. What a designer draws in it becomes a manufacturing instruction, the model goes to a factory floor and comes back as physical walls, casework and services. There is no drafting stage where being roughly right is acceptable, and no reviewer downstream who will catch a plausible-looking mistake.
That single fact governs the UX. Every interaction has to preserve engineering constraints while still feeling like design software, because the people using it are designers and sales teams working with clients in the room, not CAD operators working alone.
The same model then has to be legible to a client who has never read a floor plan. So it renders as an interactive 3D space, and as a VR walkthrough, the same data, three ways of understanding it, one of them standing inside it.
SCOPE
What I worked on across the pipeline.
- Workflows spanning 2D drafting, interactive 3D and VR against a single constraint-bound model
- UX for the VR experience, inventing interaction patterns in a domain that had none to borrow
- Custom hardware-specification workflows, where an unclear interface becomes a wrongly manufactured part
- Integration with external CAD tools including AutoCAD and Revit
- Product analytics used to decide which usability problems were worth fixing first
KEY DECISIONS
SPECIFY IN THE 3D VIEW, NOT IN A FORM BESIDE IT
Tension
Hardware specification was where errors and returns originated. It was expressed as form fields separated from the object being specified, so getting it right required holding the model in your head while reading a dropdown.
Decision
Moved specification onto the object in context, so the thing being configured and the thing being looked at are the same thing, and an invalid combination is visible rather than merely rejected.
Tradeoff
In-context configuration is far more expensive to build and to keep consistent than a properties panel. The justification was never interface quality, it was that the mistake this prevents arrives as a physical object on a truck.
VR AS A REVIEW SURFACE, NOT AN AUTHORING ONE
Tension
The obvious ambition with VR was to let people design in it. Precision work in a headset is slow, tiring, and hostile to the accuracy the model depends on.
Decision
Designed VR for the job it is genuinely better at, understanding and agreeing on a space, and kept authoring in 2D and 3D where precision lives.
Tradeoff
Less impressive as a demo. Considerably more useful in a client meeting, which is where the software actually had to win.
OUTCOME
The specification and CAD-integration work reduced backlog and returns and made the path from drawing to manufacturing measurably smoother. I am deliberately not attaching a percentage to that: the figure I was once given for it is not one I can source, so it stays qualitative.
For scale, DIRTT's own published figures give a sense of the install base the software serves.
CONTEXT, SCALE
Company figures published by DIRTT, included for context rather than as outcomes of my work.
- ICE Edge clients using the software
- 60+ICE Edge clientsCompany figure · dirtt.com/about
- active ICE licenses in the field
- 4,000+active ICE licensesCompany figure · dirtt.com/about