rakesh retweeted
one reason nvidia has been so hard to compete with is the amount of software work needed to get models running well on other hardware..
if agents can now do a good chunk of that work in 36 hours... that's a pretty big deal
still a long way from proving it works at scale.. but I'm curious how much of their advantage will hold up as agents get better at this
if agents can now do a good chunk of that work in 36 hours... that's a pretty big deal
still a long way from proving it works at scale.. but I'm curious how much of their advantage will hold up as agents get better at this
Devin spent 36 hours and ~$2k getting a model running cheaply on Trainium, work that used to take specialist teams months.
Serving AI on more available hardware is becoming more practical now that agents can cheaply automate porting and optimization work that use to require specialist teams. ⚡
0
1
0
71
rakesh retweeted
Konvene started as something else entirely..
last week it was a webcam app that checks in on how you're feeling... it got too depressing to build.. so I dropped it
then it became a multiplayer party game you play with your face.. expression controls were too fiddly and voice controls too loud.. dropped that too
what survived was the face tech.. so now it's a video call app where blur, lighting, touch-up, captions and noise reduction all run on your own device
calls go browser to browser, encrypted, nothing stored
three names later.... it's live, free and open source :)
konvene.app
last week it was a webcam app that checks in on how you're feeling... it got too depressing to build.. so I dropped it
then it became a multiplayer party game you play with your face.. expression controls were too fiddly and voice controls too loud.. dropped that too
what survived was the face tech.. so now it's a video call app where blur, lighting, touch-up, captions and noise reduction all run on your own device
calls go browser to browser, encrypted, nothing stored
three names later.... it's live, free and open source :)
konvene.app
0
1
0
58
so i was using claude code to integrate google reCAPTCHA v3 into a form... you know.. to detect bots
claude finishes the integration and proudly reports:
"i submitted the form and everything works fine"
hold on
WHO submitted the form? 😭
the bot that implemented my bot assuring me that it successfully got through my bot detection system...
i don't even know which one of us is being tested anymore lol
claude finishes the integration and proudly reports:
"i submitted the form and everything works fine"
hold on
WHO submitted the form? 😭
the bot that implemented my bot assuring me that it successfully got through my bot detection system...
i don't even know which one of us is being tested anymore lol
3
0
1
71
@SublimeCoderX Ha, v3 never blocks anything, it just hands back a score. Worth checking the server actually calls siteverify and rejects when the score is under 0.5 or the action name doesn't match, otherwise any headless browser 'works fine' too.
1
0
2
21
@prasanth_kuna server-side verification is what actually matters here.... i just found it hilarious that the first thing my agent did after implementing bot detection was proudly tell me it could submit the form... 😂
0
0
0
9
here it is: konvene.app
I'm working on a open-source and feature packed P2P video meeting service.
use my source code and enhance on it and ditch all your video meet subscriptions... please.. for god's sake.
will release it tomorrow... we have been using for months now.. not sure how services like google meet and zoom still survive in the AI era.
0
0
0
18
rakesh retweeted
🚨 HOLY CRAP! Dashcam video from just FEET AWAY shows the moment a Learjet 60 crashed in Palm Beach County, Florida
These folks are SUPER lucky to be alive.
Just INSANE.
These folks are SUPER lucky to be alive.
Just INSANE.
🚨 JUST IN: HORRIFYING new video shows what’s reportedly a Learjet NOSEDIVING into the ground in Palm Beach County, Florida
This is nightmare fuel.
No word on the number of souls on board, but obviously, there’s practically zero chance of survival here.
869
2,265
13,316
2,849,721
Konvene started as something else entirely..
last week it was a webcam app that checks in on how you're feeling... it got too depressing to build.. so I dropped it
then it became a multiplayer party game you play with your face.. expression controls were too fiddly and voice controls too loud.. dropped that too
what survived was the face tech.. so now it's a video call app where blur, lighting, touch-up, captions and noise reduction all run on your own device
calls go browser to browser, encrypted, nothing stored
three names later.... it's live, free and open source :)
konvene.app
last week it was a webcam app that checks in on how you're feeling... it got too depressing to build.. so I dropped it
then it became a multiplayer party game you play with your face.. expression controls were too fiddly and voice controls too loud.. dropped that too
what survived was the face tech.. so now it's a video call app where blur, lighting, touch-up, captions and noise reduction all run on your own device
calls go browser to browser, encrypted, nothing stored
three names later.... it's live, free and open source :)
konvene.app
0
1
0
58
rakesh retweeted
Weekend hack: HackerTube. I was annoyed with YouTube's recommendations so I had Claude filter down 43M comments over 20 years of HN to 6,454 videos that keep being recommended. *Significantly* better viewing experience, ~2 hours with Opus 5.5:
hackertube.org
- Excellent (state is kept in local browser session)
- Thumbnail is a representative frame from middle of the video instead of the super duper A/B test hyper-optimized user uploads
- Underlying data from the comment analysis as one zip file, point your Claude at it and ask for your own version
- HackerTube Longs ✊
A bit more on the process:
I use Claude Code (desktop app for Mac), Opus 5.5 on Medium effort as my default. I only talk to Claude and have it do everything, the only "manual work" I helped with in tandem was in the browser (reminiscent of MenuGen): 1) I had to create some kind of API key from Google Cloud UI to query YouTube for video metadata, 2) I booted up a VPS box for serving and 3) I bought the domain name. Once the box was up I gave Claude the IP and told it update, secure, harden, deploy and optimize the code, and it did the whole thing.
Many awesome projects that would have normally taken weeks/months even can now be few-shotted for breakfast. What else are we not thinking of.
hackertube.org
- Excellent (state is kept in local browser session)
- Thumbnail is a representative frame from middle of the video instead of the super duper A/B test hyper-optimized user uploads
- Underlying data from the comment analysis as one zip file, point your Claude at it and ask for your own version
- HackerTube Longs ✊
A bit more on the process:
I use Claude Code (desktop app for Mac), Opus 5.5 on Medium effort as my default. I only talk to Claude and have it do everything, the only "manual work" I helped with in tandem was in the browser (reminiscent of MenuGen): 1) I had to create some kind of API key from Google Cloud UI to query YouTube for video metadata, 2) I booted up a VPS box for serving and 3) I bought the domain name. Once the box was up I gave Claude the IP and told it update, secure, harden, deploy and optimize the code, and it did the whole thing.
Many awesome projects that would have normally taken weeks/months even can now be few-shotted for breakfast. What else are we not thinking of.
1,952
4,203
46,322
4,197,927
the "own your intelligence" discussion has me thinking about something else...
we spent years letting algorithms decide what appeared in our feeds
now we're building AI assistants that could influence what we read, buy, research and even believe
i really hope we don't hand over that much control to a handful of companies again....
we spent years letting algorithms decide what appeared in our feeds
now we're building AI assistants that could influence what we read, buy, research and even believe
i really hope we don't hand over that much control to a handful of companies again....
0
0
0
67
rakesh retweeted
Own your intelligence.
Anyone that opposes this has something to sell you (perhaps your own intelligence back to you.)
Local models need to continue proliferating. The “absolute unit” massive models may be out of reach for now locally, so your best option is to, indeed, keep your options open. Own your harness, data, connections, runtime: your agents. Embrace open source.
Anyone that opposes this has something to sell you (perhaps your own intelligence back to you.)
Local models need to continue proliferating. The “absolute unit” massive models may be out of reach for now locally, so your best option is to, indeed, keep your options open. Own your harness, data, connections, runtime: your agents. Embrace open source.
352
351
3,360
206,963
one reason nvidia has been so hard to compete with is the amount of software work needed to get models running well on other hardware..
if agents can now do a good chunk of that work in 36 hours... that's a pretty big deal
still a long way from proving it works at scale.. but I'm curious how much of their advantage will hold up as agents get better at this
if agents can now do a good chunk of that work in 36 hours... that's a pretty big deal
still a long way from proving it works at scale.. but I'm curious how much of their advantage will hold up as agents get better at this
Devin spent 36 hours and ~$2k getting a model running cheaply on Trainium, work that used to take specialist teams months.
Serving AI on more available hardware is becoming more practical now that agents can cheaply automate porting and optimization work that use to require specialist teams. ⚡
0
1
0
71
My vlog today was fully edited by Claude Code with Opus 5.5!
I copied the video clips from my DJI to a folder, then opened Claude Code in it and asked it to edit it into a 10 minute vlog
I made a few adjustments but not a lot
It transcribed everything I said with mlx-whisper (Whisper large-v3-turbo running on my MacBook Pro's GPU through Apple's MLX)
It pulled video frames every few seconds into grid images (contact sheets), so Claude could see what's in each clip
It used SpeechBrain ECAPA (a voice-matching model) to figure out what was my voice from other people
It then made thumbnails with OpenCV with YuNet's face detector
The edit itself is a list of cuts in a JSON file: which clip, start time, end time, Claude then picked the pieces by reading the transcripts and looking at the frame grids by itself
I added on screen captions for X and on YouTube for the first minute with Pillow (Python's image library)
P.S. I didn't ask for any of this, and I didn't know any of this until @thesamparr just asked me how I edited it
All I asked for was "Claude can you edit these clips into a 10 minute vlog for X and YouTube"
I think if I didn't have AI to edit it, I wouldn't be vlogging because editing would be such a painful and slow process for me and hiring someone to do it would be even more hassle for me!
I copied the video clips from my DJI to a folder, then opened Claude Code in it and asked it to edit it into a 10 minute vlog
I made a few adjustments but not a lot
It transcribed everything I said with mlx-whisper (Whisper large-v3-turbo running on my MacBook Pro's GPU through Apple's MLX)
It pulled video frames every few seconds into grid images (contact sheets), so Claude could see what's in each clip
It used SpeechBrain ECAPA (a voice-matching model) to figure out what was my voice from other people
It then made thumbnails with OpenCV with YuNet's face detector
The edit itself is a list of cuts in a JSON file: which clip, start time, end time, Claude then picked the pieces by reading the transcripts and looking at the frame grids by itself
I added on screen captions for X and on YouTube for the first minute with Pillow (Python's image library)
P.S. I didn't ask for any of this, and I didn't know any of this until @thesamparr just asked me how I edited it
All I asked for was "Claude can you edit these clips into a 10 minute vlog for X and YouTube"
I think if I didn't have AI to edit it, I wouldn't be vlogging because editing would be such a painful and slow process for me and hiring someone to do it would be even more hassle for me!
107
78
2,039
184,036
i've been thinking... if AI makes building software cheap enough, will we still spend years adapting to someone else's app?
or will we just build our own?
maybe the next frontier in software isn't about reaching a million users... it's about making something useful for just one.
or will we just build our own?
maybe the next frontier in software isn't about reaching a million users... it's about making something useful for just one.
0
0
0
29
love how every other server / website is getting hacked.
covid 2.0 indeed.
brace yourselves... we are heading towards a temporary shutdown which will force everyone to fix their security vulnerabilities..
and then all the security startups (which was built around the AI stuff) will go out of business.
wonderful times indeed..
covid 2.0 indeed.
brace yourselves... we are heading towards a temporary shutdown which will force everyone to fix their security vulnerabilities..
and then all the security startups (which was built around the AI stuff) will go out of business.
wonderful times indeed..
0
0
0
65