@AviAisenberg@Nominal_io I’m also getting a “Something went wrong. Please try again.” With no feedback on what’s wrong with the form. I’m on iOS mobile.
@zachglabman@amypretzel@connorkapoor@FilArons Feels like you don't really retrofit it. Kind of like adding git to an old codebase, nobody goes back & rewrites history, you just start committing from today. Tag every build & test run going forward & let agents backfill the old stuff when someone actually needs it.
@maketaster@bscholl Feels like the missing piece in Blake's post is test. Once the algorithm designs the part, the test data has to come back & tell you what it got wrong. Same thing we're going through in software right now.
@bscholl@jcm10x1 Kind of what's happening in software right now with software factories. AI writes the code, while the hard part is the checks, verification, and validation. Feels like hardware's next step is test data flowing back & checking the algorithm on its own.
@JasBilyeu Kind of what's happening in software right now. AI writes most of the code & engineers build the checks. The loops will get tighter and tighter. Building connected physical verification tools seems like a good place to move to.
@JasBilyeu Say AI spits out 10 bracket designs overnight. The ME's job is defining what "good" means (loads, tolerances, how we'll test it), then wiring up the checks: sim, scan, load test, all flowing back on their own via some AI agent. Then you only review what fails & tighten the spec.
@MengTo I want this... for the Apple Vision Pro, with multiple player with different angles, using voice, with a way to pin the "table" to actual physical objects (like a table). Incredible!
@zachglabman Whole heartedly agree, but it’s more showing the need is rising and rising that even Silicon Valley notices the need. It may not be the right place for it to sprout from, but the gap is apparent, and the demand clear.