
Co-founder & CTO @neosapienHQ. Building Neo, the wearable that remembers. Buddhist practice. Reads sci-fi as product research. @iitmadras. Forbes 30u30 Asia
Bengaluru, India
Joined February 2015
- Tweets645
- Following394
- Followers545
- Likes1.4K
Pinned Tweet
We have big retention problem.
It’s growing too much 😏
We have a BIG retention problem.
Our retention has been increasing month over month, far beyond what we expected.
We just crossed 500K minutes of user conversations in a single day, taking us to 37M+ minutes of conversations to date.
The challenge? The more users come back, the more compute we need.
We are doubling down on cost optimisation and believe open-source models will continue to drive inference costs down.
When users keep coming back, you know you’re building something that has a strong PMF.
we don't really look at stuff anymore. we just consume it, like we're always a little hungry
scroll, next scroll. a reel, a podcast at 2x. even the good stuff goes in and out without landing anywhere. we finish a book and can't remember a single line a week later
the weird part is it's real hunger. what our eyes and ears take in is food too, it just never fills us up
a detox doesn't fix it. we stay off for a week and come back just as hungry
what helps is a small behaviour shift. that half second before the thumb moves again, just notice it. not fight it, not feel bad about it. just catch ourselves reaching
turns out the moment we actually look, we're not that hungry anymore
3
71
Finally a movie where the board meeting is the action scene
New trailer for Luca Guadagnino's ‘ARTIFICIAL’, starring Andrew Garfield as Sam Altman.
The film follows the story involving OpenAI focused on the firing & rehiring of CEO Sam Altman.
In theaters on January 8.
Artificial - Official Trailer - In Select Theaters Christmas
1
132
Neuroscientists spent years scanning brains looking for the self
Monks found the same thing sitting on a floor with their eyes closed
Sit in silence my friends :)
A meta-analysis of 912 neuroimaging studies finds no single core brain region or network underlying the self, challenging the prevailing view of a shared cortical midline or insular anchor. Instead, self-related processing emerges from contextually determined, shifting coalitions of distributed brain systems, suggesting that selfhood is defined by dynamic configuration rather than a fixed neural substrate.
biorxiv.org/content/10.64898…
76
Meditation retreats about to get a leaderboard
Introducing AIR-O, a model that decodes what a person is attending to directly from brain scans.
Using only noninvasive brain data, we can generate an image from scratch in real time that depicts where a person is focusing their attention.
This is the first of a new generation of algorithms that can read the shape of thoughts directly from brain activity.
3
132
GTA VI drops tomorrow
attention is one stream
feed, AI tools, work, people
all from the same few hours a day
If you are shipping this month, your competitor is gta 6
192
Hardware is a moat because nobody wants to do hardware :)
so many founders I talk to don’t want to build software products bc they’re scared it’s all just going to get eaten up. so they’re moving into hardware (even if they have no background in it) to build something with a defensible moat. and if this trend continues our physical world is going to look so radically different soon
5
265
Good timing
'Human' is Cambridge Dictionary's word of the year.
"Human", meaning to have the "qualities, faults, and feelings that people have, as opposed to gods, animals, or machines"
u.afp.com/SFWS
56
genuinely what is Yann LeCun yapping about
the guy invented half of what we use and massive respect to that
but some of his talks just doesn't make sense
"they can't reason or plan" while agents went from 4 minute tasks to 12 hour ones. you don't finish a 12 hour task without planning
Some delusion is fine but what the actual fuck ?
2
136
The thing that helps me most with founder anxiety is feeling my whole body at once
Anxiety mostly lives in thought. Your mind is replaying a call or running the next one
But attention is limited. When it's busy feeling the whole body, there's less of it left for the story. The loop doesn't get fed and it quiets down
You can do it anywhere:
1. Eyes open, keep doing what you're doing
2. Feel your feet on the ground
3. Let it spread up your legs
4. Then your belly, chest and back
5. Arms, hands, neck and face
6. Now all of it at once
Works in a cab, a queue or before a meeting
1
9
359
if you build hardware in India an afternoon with the person who ran Xiaomi here is about as good as it gets
thanks for coming by @manukumarjain
It was great hosting @manukumarjain at our office. He is a veteran when it comes to understanding hardware, and his journey at Xiaomi has been inspirational. It was the largest consumer company effectively overtaking HUL (Hindustan Unilever) at that time.
closed loop self improvement hasn't started yet. no model is retraining itself overnight
but METR estimates Anthropic's progress has been about 1.9x faster since Oct 2025, mostly from AI helping with the engineering
so the loop is probably already half closed wherever the result is quick to check
65
RSI so far works wherever a test can score the result in seconds. kernels, configs, speedruns
the real step is a model deciding what's worth working on. I haven't seen that yet
Everytime I see a Dwarkesh video, I'm wondering what they are smoking in SF. It seems very distant from reality
"Superintelligence" is one of the cringiest terms I've seen, it's a rebrand of AI to get more money out of VCs.
An no, recursive self-improvement hasn't started. It's just human researchers prompting to say "speed it up". It does some cute kernel optimizations and hyperparameter sweeps, very similar to the NanoGPT speed run records done by humans. Nothing fundamental or groundbreaking gets invented by so-called superintelligence.
To give an example: when @Recursive_SI released their (impressive) Flint model on @huggingface, I cringed a little for their use of the terms "superintelligence" + "recursive self-improvement". As said in their blog post, through "recursive SI", LLMs were able to figure out - I kid you not - to determine how many total parameters and active parameters the model should have. So yeah, it's just changing a config. It's doing hyperparameter sweeps on things humans have built (like MoE's, validation loss, perplexity etc). Of course, you shouldn't tell that to VCs. You should scare them by saying AI is improving itself at an alarming rate - this increases the $$$ raised by another factor.
Can we please stop all this misinformation: no, superintelligence is not here. No, recursive self-improvement hasn't started. And yes, there's lots of work to do to build more reliable AI that is actually smart.
This is why there's such a huge gap between super impressive benchmark numbers of LLMs, and the (lacking) economical impact of them. It's not able to do research like a human, doing experiments, coming up with actual new stuff, and generalize beyond their large amounts of training data.
1
1
113
Anthropic put "solely for illustrative purposes" under the IPO chart
Every seed deck hockey stick in Bengaluru has been illustrative too :)
This slide from the Anthropic IPO roadshow was a big factor in my decision to leave. It’s an insane plot and I’m surprised they made it public.
It shows the compute efficiency of pretraining over time. The y-axis is a **log scale**: they are seeing continued superexponential algorithmic progress.
Mythos was not a one-time deviation, but a point on a superexponential. They could already retrain Mythos with a fraction of the compute originally needed.
It doesn’t strictly imply a software-only intelligence explosion- they have been spending more on R+D. But algorithmic improvements do not require ridiculous creative genius and are on the verge of being automated. You do the math.
2
138























