A prosthetic socket is shaped by hand over weeks of visits, and if pressure at the tip of the residual limb is too high, the skin breaks down and the prosthesis goes unused.

A web app for the prosthetist: enter the patient's weight and limb measurements, and it returns the socket liner design that relieves the limb tip the most.

The study they built on reports around 30 minutes for a single simulation; their app runs 216 full 3D simulations in parallel in about 90 seconds.

Running the whole design sweep at once is what makes the tool usable inside a patient visit rather than as an overnight job.

Their original idea was cold chain transport for vaccines and insulin; after talking to our team they reframed the search around computationally heavy optimization problems.

Of a six-hour window, roughly half went into understanding the problem and the model's limits before any simulation was run.

With a clear problem and a clear goal, their work was mostly in framing the right prompts; the physics and the geometry script came out of AI coding tools.

Their 90-second target was self-imposed for the demo, and the next step would be a far larger parameter sweep over the design space.