Hardware prototyping just got a shorter feedback loop. A user named jordi8 posted on the OpenAI community forum about using Codex, inside ChatGPT with no terminal required, to design and physically produce a 3D printed enclosure with working mechanisms.
What Happened
Jordi described the object in natural language. Codex generated the OpenSCAD code and the STL file. The enclosure included a sliding lid, a flexible mount, and magnet integration. At some point, jordi gave Codex access to his computer and it began sending jobs directly to the printer on its own.
The project did not stop at the enclosure. Codex also wrote the code running on the Raspberry Pi inside the device and produced a visual wiring guide showing exactly where to connect each pin. The finished device connects to the internet and uploads data to an app that Codex also built.
Why This Matters for Operators
The barrier to custom hardware has always been the gap between an idea and the files a machine can execute. Knowing OpenSCAD, managing a slicer, and wiring a microcontroller correctly are three separate skill sets. This workflow compresses all three into a description.
It is still early and jordi’s post is a single anecdote, not a benchmark. But the fact that an OpenAI community manager (Fede_s) responded with an invite to a physical builds program suggests the company is paying attention to this use case.
