This is great! And you've done the hard part super well. You know, the actual underpinning tech.
But the "easy" part— I would immediately bounce off your page due to its clear vibe-designed and, most importanty, -worded nature.
Eyebrows like "Newer demos · newest first", italicized and contrastly colored words in sentences, and the two-to-three sentence phrased paragraphs (like "Your best workers do the task wearing our EMG sleeve, on camera, with hand tracking. The sleeve leaves the hand completely unobstructed.") absolutely SCREAM AI-generated.
Just replace it with even a mediocre version of it that you write yourselves and I think you'd do yourselves a ton of favors.
pmcarlton 1 hours ago [-]
"The hand disappears. The sleeve still knows." yep I'm outta here.
(Turned off, that is; I'm very impressed by the tech.)
knuckleheads 59 minutes ago [-]
I've wanted these for measuring how much force is applied via the hands during partner dancing for a long time. There's a lot of subtly there that isn't visible and I've long wondered how to measure and visualize that. Very neat! Hope I get to try one someday.
millspore 3 hours ago [-]
How does it handle fatigue shift, like if someone's been gripping stuff for an hour the emg baseline must drift pretty hard
Benard-dev 3 hours ago [-]
Interesting approach. Been seeing more AI tools that can decompile and annotate binaries almost instantly now — makes old obfuscation feel useless.
How are you handling false positives when hardening? Does it break legit reversing tools like Ghidra or just slow down LLM-assisted ones?
Curious to try it on a small Go binary I have.
12_throw_away 5 minutes ago [-]
ha, this comment would actually be plausible on like 95% of HN submissions
unpleasurable 2 hours ago [-]
Are you planning to control human arms with this?
3 hours ago [-]
cloakley 3 hours ago [-]
I would quit my job if they forced me to wear that glove to track what I do.
Onavo 3 hours ago [-]
People in India already wear head mounted cameras purely for data collection. You can look up the videos online. Where do you think companies like Scale and Surge gets their data? World models can't happen ex nihilo..
Rendered at 00:00:52 GMT+0000 (Coordinated Universal Time) with Vercel.
But the "easy" part— I would immediately bounce off your page due to its clear vibe-designed and, most importanty, -worded nature.
Eyebrows like "Newer demos · newest first", italicized and contrastly colored words in sentences, and the two-to-three sentence phrased paragraphs (like "Your best workers do the task wearing our EMG sleeve, on camera, with hand tracking. The sleeve leaves the hand completely unobstructed.") absolutely SCREAM AI-generated.
Just replace it with even a mediocre version of it that you write yourselves and I think you'd do yourselves a ton of favors.
(Turned off, that is; I'm very impressed by the tech.)
How are you handling false positives when hardening? Does it break legit reversing tools like Ghidra or just slow down LLM-assisted ones?
Curious to try it on a small Go binary I have.