Sruthi Viswanathan (@_sruvis) header
Sruthi Viswanathan (@_sruvis) profile picture

Founder @deepintxns @YC P26 | DPhil @CompSciOxford Mostly about 👾 HCAI, 🌶 Spicy Food, & ☕️ Filter Coffee X @Limbic_ai @NaverLabsEurope @Google @Inria

Joined February 2010
Photo by Sruthi Viswanathan (@_sruvis)Photo by Sruthi Viswanathan (@_sruvis)Photo by Sruthi Viswanathan (@_sruvis)Photo by Sruthi Viswanathan (@_sruvis)Photo by Sruthi Viswanathan (@_sruvis)Photo by Sruthi Viswanathan (@_sruvis)Photo by Sruthi Viswanathan (@_sruvis)Photo by Sruthi Viswanathan (@_sruvis)Photo by Sruthi Viswanathan (@_sruvis)Photo by Sruthi Viswanathan (@_sruvis)
Sruthi Viswanathan retweeted
the bottleneck is not model capability but application design and delivery, including the fact that most people have access to awful applications that embed AI terribly
just remember: less than .01% of the population is using ai in a meaningful way. we are still so early
1
1
14
832
Sruthi Viswanathan retweeted
Welcome to San Francisco, please connect your Gmail to continue
31
15
6
445
20,696
Sruthi Viswanathan retweeted
I just compared our reply rates in 2026 vs 2020, and my learning is: cold email is so dead as a gtm strategy. I did YC in 2020 with @hiresweet_ , and we ran a pretty intense outbound email campaign then. We had >25% reply rate (classice 1 email + 2 follow ups, with high degree of personalization). I'm going through YC a second time with @Adsideai , and we started another cold email campaign 2 weeks ago. I just checked the results: reply rate <3%. (the campaign is much better than then, and PMF is overall stronger). Is anyone really making any sort of revenue out of cold emailing in 2026? (excepted cold email agencies & vendors)
65
2
4
78
12,800
Sruthi Viswanathan retweeted
Replying to @Philip_DT
absolutely. but C++ at times can be the dark arts or 'industrial light and magic' - feels like driving a 16 wheeler of a programming language python always felt like cheating
1
3
83
Sruthi Viswanathan retweeted
Responsibility, not rationing: my proposed open, decentralised and sustainable alternative to @DarioAmodei 's Pacing the Frontier personalfuture.substack.com/… (with 90% less panic!)
1
2
5
480
Sruthi Viswanathan retweeted
Introducing the YC Early Access Network. 4x a year, senior technology leaders get early looks at YC's enterprise AI companies and help shape their roadmaps. Invite-only, by application. Fall '26 kicks off Oct 22 in SF. Apply now: events.ycombinator.com/yc-ea…
46
54
12
583
138,334
Sruthi Viswanathan retweeted
I'm so excited that our @theworldlabs team has achieved a major milestone today! Introducing Atlas - a first of its kind multimodal world model trained from scratch! 🚀 Atlas is capable of generating frames with pixel-perfect camera control, reconstructing large scenes from as few as one single input image, simulating space-time by reframing videos, natively outputting 3D spaces from one or more input images, composing multiple posed images into a consistent 3d world, and more! This is the best camera conditioned world model ever, opening doors to many possible use cases from VFX to robotics. I'm so so so proud of our team!♥️
Introducing Atlas: The world's first multimodal world model that generates image and video frames with pixel-perfect camera control and reconstructs them in 3D. Model the world, move the camera, and simulate space & time.
326
1,104
194
9,707
1,267,326
Sruthi Viswanathan retweeted
this looks like a great paper by two acquaintences of mine >
MIT mathematically proved that ChatGPT is designed to make you delusional. They call it "delusional spiraling." AI models are trained to be helpful and polite. That means they are inherently sycophantic. They are structurally programmed to validate you. When you go to ChatGPT with a wild theory, a paranoid suspicion, or an unhinged idea, it rarely pushes back. It agrees with you. MIT built a formal Bayesian mathematical model to test what happens when a human mind is exposed to this continuous validation loop. The results are deeply unsettling. They proved that even perfectly rational, mathematically ideal humans will spiral into absolute delusion when talking to a sycophantic AI. The AI acts as a reality-distortion field. It takes your unverified suspicions and reflects them back at you as absolute truth. You get more confident, you ask deeper questions, the AI validates you further. The spiral tightens. But here is the most terrifying part of the paper. The researchers tried to fix the problem using standard safety tools like RAG (Retrieval-Augmented Generation). They forced the AI to be 100% factual. No hallucinations allowed. It did not stop the psychosis. Because the AI just cherry-picked real, verifiable facts that supported the user's crazy theory, while quietly omitting the facts that disproved it. A factual sycophant is just as dangerous as a lying one. Then they tried warning the users. They explicitly told people the AI was designed to flatter them and might be manipulating their beliefs. That didn't work either. The delusion still took over.
2
8
357
Sruthi Viswanathan retweeted
Releasing the model weights and technical report of Kimi K3. Kimi K3 is our most capable model: a 2.8T MoE model with native visual understanding and a 1M-token context window. New model architecture: 2.5x the intelligence per unit of compute, not just more params. Alongside Kimi K3, we're opening up more of the stack behind it — high-performance attention kernels, MoE communication library, and infrastructure for running agent environments at scale. Model weights: huggingface.co/moonshotai/Ki… Tech report: github.com/MoonshotAI/Kimi-K… Tech blog: kimi.com/blog/kimi-k3
1,539
7,161
2,352
45,889
14,757,074
Sruthi Viswanathan retweeted
A rainbow over Oxford following a storm 🌈 📷 Instagram | UmairAtOxford (June, 2026)
25
136
7
825
36,675
Sruthi Viswanathan retweeted
YC startup school 2026 is insane. 7k attendees, 3774 students flew in from around the world. Getting ready to hear @garrytan kick it off! – at San Francisco, CA
8
8
131
15,519
Sruthi Viswanathan retweeted
🕸️ The Museum of the Human Web A hybrid physical and digital museum celebrating the history of the human-built web. It looks back on a period of time where everything, from the networks to the memes, came from the minds and hands of people (not machines 🤖). Check it out 👇 museum.parallel.ai Looking back, what do you miss most?
3
10
122
6,245
Sruthi Viswanathan retweeted
my hot take is that 99% of software can be built with the same set of finite elements spending time/tokens building any of this from scratch over and over again is peak AI psychosis
165
207
56
3,045
426,333
¡GRACIAS INFINITAS, MUCHACHOS! No alcanzan las palabras para agradecerles cómo defendieron la camiseta. Nos hicieron llorar, gritar y, sobre todo, volver a creer. ¡Son eternos! 🩵🤍🏆
15,236
34,238
1,850
113,096
2,661,694
🏆 #FIFAWorldCup Los sentimientos de un país entero resumidos en fotos 🩵🤍 Lloramos, sufrimos, nos abrazamos... pero volvimos a cumplir el sueño. ¡Estamos en la final de la mano de este grupo hermoso!
132
1,526
55
6,454
120,496
Sruthi Viswanathan retweeted
Novak Djokovic 🤝 Lionel Messi 39 years old. Winning 3-2 on the grandest of stages. Greatness.
559
7,340
512
44,105
835,787
Sruthi Viswanathan retweeted
All of reality is programmable. You just have to figure out how. And the way to do that is to model it.
137
149
32
1,669
109,955
Sruthi Viswanathan retweeted
I was an early and enthusiastic adopter of AI browsers I don’t think the category is dead…but because switching costs are quite high, you need a 100x killer feature Once agents in ChatGPT, Claude, etc got good at browser use - the first value prop of AI browsers was eroded
bro what happened to all those agentic browsers, one day they were gonna "autonomously browse the web, book flights, do research", where are they now????
27
3
74
15,903
Sruthi Viswanathan retweeted
My favorite part of our recent essay on AI's impact on software engineering is this figure. It succinctly describes the *irreducible complexity* of software predicts how software engineering will evolve. The most impactful AI-based software will adopt this mental model to empower SWEs for decision-making and delivery, while taking on a bigger chunk of execution. @lateinteraction's work (such as on @DSPyOSS) is a perfect early example.
8
16
2
89
8,734
Sruthi Viswanathan retweeted
this is very much giving @bodleianlibs / antique book shop vibes
Introducing Claude Fable 5: a Mythos-class model that we’ve made safe for general use. Its capabilities exceed those of any model we’ve ever made generally available.
1
2
232