AI can produce an impressive site plan image in seconds. But can it research a real property, create useful CAD linework, and help develop a concept in Civil 3D?
I put ChatGPT through that workflow to find out. Starting with a property screenshot, I asked it to research the site, assemble a preliminary Site Investigation Report (SIR), explore possible land uses, and export DXF files. Then I opened those files in Civil 3D to see what I actually had.
From screenshot to site research
The first test was whether AI could use the screenshot to find GIS property information and begin identifying the site’s constraints. I also asked it to create an SIR covering items such as zoning and setbacks.
That gave me a starting point for investigation. It did not settle the design questions. Parcel data needs to be checked against the GIS source, and code requirements need to be confirmed with the governing jurisdiction before they guide a real project.
Getting property lines into Civil 3D
Next, I had ChatGPT export property boundary linework as a DXF. I imported it into Civil 3D and worked through the coordinate system and geolocation so I could compare the lines with the map.
This is where a convincing image meets a measurable drawing. The linework has to land in the right place. For preliminary exploration, GIS data can be useful; precise property boundaries still require appropriate survey information.
Testing the site plan concept
I then asked AI to suggest potential land uses and generate a conceptual site layout. It produced a plan image and a DXF export, which I brought into CAD for a closer look.
The import is the revealing part of the experiment. In an image, a layout can look complete. In Civil 3D, I can inspect the layers and individual lines, measure the geometry, and see what would need to be rebuilt or developed further. A concept only becomes useful when its access, parking, circulation, setbacks, grading, drainage, and other constraints can hold up to that scrutiny.
So, are civil design jobs safe?
This test shows why the answer is more complicated than a dramatic headline. AI can accelerate research and generate early options. It can even produce linework that opens in CAD. Those are useful capabilities.
But producing a file is only one part of civil design. Someone still has to verify the inputs, spot conflicts, apply the governing requirements, and turn a rough concept into a coordinated plan. For now, I see AI changing the starting point of the work more clearly than replacing the judgment needed to finish it.
Watch the video to see the actual DXF imports and decide how useful the results would be in your own workflow.
