Turning companies AI native | IITM '22 Prev: @alma_inc, @JioHotstar, Unacademy

Talk to me ➡️
Joined March 2016
Replying to @amrnth0
Folks new here: I'm helping companies transform their work with AI. If you think there's repetitive work that you want to get automated, or you are exploring it in general, I'd love to speak. Find 20 mins in my calendar here: calendar.app.google/wezWE24a…
6
3,909
Yes please!
Time to go after aging - the most important problem of the milenium
1
41
Wondering what Peter Thiel would say about Burma Burma's $50M valuation
6
218
Anthropic's proposed spends on compute is bonkers! For example, it'll be paying $3.1B/mo to SpaceXAI from Dec 2026! Also there's a committment to pay Google upto $114.5B till 2033(plus $125B-$150B estimated) Source: sec.gov/Archives/edgar/data/…
3
133
No-shows on scheduled calls hurt a bit. The logical response to this should be to move on with it as this is expected in a numbers' game. But the weak human in me takes this personally. I'm not worried as with repetition it'll be stronger.
1
18
485
For any government work, my best buddy is ChatGPT's browser use. I filed my Income Tax Returns this year myself thanks to it. I withdrew my accumulated EPF also using it. Without it, it would have been a nightmare to navigate government websites. Everybody should use a Browser Use app to navigate messy websites like Indian government sites.
wow this visa application site is.....a ux marvel
13
707
He's an artist. Has a way with words.
A cunning move by Aer Lingus 😉
56
"The eyes chico... They always lie"
6
184
Whoa, it's that cheap to train such an amazing voice model huh? The data(human speech) also is abundantly available. So with iterations over the model arch, and quick human-feedback you could replicate this model. The arch iterations can be done with a frontier LLM(Recursive improvement). The bottlenecks are time and money($10/hr for the GPU)
Yooo this full duplex realtime voice actually slaps, trained on only a single GB200 NVL72. We need more realtime voice competition outside OAI, this is exciting!
3
273
Wow! Game changer. If this supports Hindi it'll be amazing.
We believe voice is the most natural way to interact with intelligence, so we created Dessa, the most authentic full-duplex speech model ever built. Talking with it feels like talking with a person…full of emotion, nuance, pauses, and a natural cadence. block.xyz/dessa
1
158
Pretty excited about this one. We just received a $30,000 in-kind grant from Anthropic to support our startup working on clinical trial simulation, and have been selected for their AI for Science program. Always a great time working with @Vinay_Neuro, and of course a lot of important work ahead. Will be posting updates here.
8
2
57
3,720
Amarnath retweeted
atp if you think you're extremely cracked, you should seriously consider taking 6mos off work and finding out what you can do with your own interests
9
3
182
7,255
Amarnath retweeted
We’re shutting down too. We don’t use php I just don’t wanna work tomorrow.
We are aware of the PHP vulnerability and are shutting down for the weekend.
79
2,260
94
35,030
516,130
Dario haters in shambles right now. He warned us long ago.
9
403
Dario proven right yet again
Not just PHP. Node.js has also announced security releases for October 21, with HIGH-severity vulnerabilities affecting its 26.x, 24.x and 22.x release lines. If you run either stack in production, keep an eye on official advisories and apply patches promptly. Security maintenance doesn't stop at your application code. Your runtime matters too. Node.js advisory: nodejs.org/en/blog/vulnerabi…
1
239
Watch out folks!
ffs, litellm compromised again! if you installed llitellm==1.107.0, good chance your environment is compromised
3
216
Was trying to incorporate a Pvt. Ltd., and apparently you have to spend INR 50k to 1L every year just for compliance 🥲
107
48
12
1,095
72,916
Disclaimer: This was revealed to me by Claude
1
45
8,005
Wow, indefinite caching is such a blessing for users.
Day 16/ You can now cache for 12h in the API and for logged in users across Codex & Work we keep cache forever and you don’t have any penalty for coming back to an old conversation or resuming after a few hours and are now guaranteed to hit the cache.
1
2
465