Arjun

Recommendation systems feel like they should know you by now. They track what you watch, when you pause, what you drop halfway, what you come back to.🔀 And still, what they suggest often feels slightly off. Not because they lack data, but because they’re solving a different problem than you are. This week's piece talks about why recommendations feel broken. Link Below!🔗

Today, India takes a defining step in its civil nuclear journey, advancing the second stage of its nuclear programme. The indigenously designed and built Prototype Fast Breeder Reactor at Kalpakkam has attained criticality. This advanced reactor, capable of producing more fuel than it consumes, reflects the depth of our scientific capability and the strength of our engineering enterprise. It is a decisive step towards harnessing our vast thorium reserves in the third stage of the programme. A proud moment for India. Congratulations to our scientists and engineers.

Delighted to be a founding member of @nvidia's Nemotron coalition. Bullish on what the coalition can achieve together. And India's sensibilities of being voice-first, linguistically diverse, and cost efficient at population-scale will be at the global table shaping frontier research!

📢 Open-sourcing the Sarvam 30B and 105B models! Trained from scratch with all data, model research and inference optimisation done in-house, these models punch above their weight in most global benchmarks plus excel in Indian languages. Get the weights at Hugging Face and AIKosh. Thanks to the good folks at SGLang for day 0 support, vLLM support coming soon. Links, benchmark scores, examples, and more in our blog - sarvam.ai/blogs/sarvam-3…

🚨 BREAKING: Someone just made 70B parameter models run on a single 4GB GPU. It's called AirLLM. No quantization. No distillation. No pruning. Just raw 70B inference on hardware that costs less than a dinner. You can even run Llama 3.1 405B on 8GB VRAM. Here's how it works: → Decomposes the model layer-by-layer → Loads only one layer into GPU memory at a time → Runs inference, moves to the next layer → Prefetches the next layer while computing the current one → Supports 4-bit and 8-bit compression for 3x speed boost No cloud API. No $10K GPU. Just pip install airllm and go. Here's the wildest part: It supports almost every major model — Llama, Qwen, Mistral, ChatGLM, Baichuan, InternLM — and it auto-detects the model type. One line of code to load. One line to generate. Works on Linux, macOS (Apple Silicon), and even Google Colab free tier. Your old gaming laptop can now run the same models that needed an A100. 100% Open Source. Apache 2.0 License.

Three New Funds for Exceptional Founders in India, APAC and beyond 🚀 We are pleased to announce the closing of $1.3 billion in new capital commitments across our India Seed, India Venture, and APAC funds. Along with significant uninvested capital in our existing Growth fund, we are excited about our ability to back outlier founders building category defining companies, across multiple stages, as we have done for two decades. We are immensely grateful for the steadfast commitment from our Limited Partners (LPs) for our inaugural Peak XV funds. Many of our LPs are leading non-profit endowments and foundations, and serving them adds a special meaning to our work. We understand every fundraise is a responsibility to the founder ecosystem and to our LPs, and we are relentlessly committed to serving them. This is the most exciting time we have witnessed in our lifetimes. AI is transforming the world at an unprecedented pace, and while the initial breakthroughs were concentrated in Silicon Valley, AI opportunities are now abundant in India and APAC. The size, scale and sophistication of technology startups is deeply inspiring across both India and APAC. India’s FinTech ecosystem is already one of the most advanced in the world and the Consumer opportunity has decades of compounding ahead. The combination of technical innovation, deepening markets, quality of talent and increasing global ambition makes this an amazing time to be investing in the region. For those special founders who have a vision to change how something works and who combine ambition with execution, hustle with resilience and who wish to serve the world ahead of themselves, we are here to partner with you. 🙌 @sjs_day1 @mobhat @gvravishankar @RajanAnandan @i_sakshichopra @rohitagar_wal @abhishek_mohan

Shifting structures in a software world dominated by AI. Some first-order reflections (TL;DR at the end): Reducing software supply chains, the return of software monoliths – When rewriting code and understanding large foreign codebases becomes cheap, the incentive to rely on deep dependency trees collapses. Writing from scratch ¹ or extracting the relevant parts from another library is far easier when you can simply ask a code agent to handle it, rather than spending countless nights diving into an unfamiliar codebase. The reasons to reduce dependencies are compelling: a smaller attack surface for supply chain threats, smaller packaged software, improved performance, and faster boot times. By leveraging the tireless stamina of LLMs, the dream of coding an entire app from bare-metal considerations all the way up is becoming realistic. End of the Lindy effect – The Lindy effect holds that things which have been around for a long time are there for good reason and will likely continue to persist. It's related to Chesterton's fence: before removing something, you should first understand why it exists, which means removal always carries a cost. But in a world where software can be developed from first principles and understood by a tireless agent, this logic weakens. Older codebases can be explored at will; long-standing software can be replaced with far less friction. A codebase can be fully rewritten in a new language. ² Legacy software can be carefully studied and updated in situations where humans would have given up long ago. The catch: unknown unknowns remain unknown. The true extent of AI's impact will hinge on whether complete coverage of testing, edge cases, and formal verification is achievable. In an AI-dominated world, formal verification isn't optional—it's essential. The case for strongly typed languages – Historically, programming language adoption has been driven largely by human psychology and social dynamics. A language's success depended on a mix of factors: individual considerations like being easy to learn and simple to write correctly; community effects like how active and welcoming a community was, which in turn shaped how fast its ecosystem would grow; and fundamental properties like provable correctness, formal verification, and striking the right balance between dynamic and static checks—between the freedom to write anything and the discipline of guarding against edge cases and attacks. As the human factor diminishes, these dynamics will shift. Less dependence on human psychology will favor strongly typed, formally verifiable and/or high performance languages.³ These are often harder for humans to learn, but they're far better suited to LLMs, which thrive on formal verification and reinforcement learning environments. Expect this to reshape which languages dominate. Economic restructuring of open source – For decades, open-source communities have been built around humans finding connection through writing, learning, and using code together. In a world where most code is written—and perhaps more importantly, read—by machines, these incentives will start to break down.⁴ Communities of AIs building libraries and codebases together will likely emerge as a replacement, but such communities will lack the fundamentally human motivations that have driven open source until now. If the future of open-source development becomes largely devoid of humans, alignment of AI models won't just matter—it will be decisive. The future of new languages – Will AI agents face the same tradeoffs we do when developing or adopting new programming languages? Expressiveness vs. simplicity, safety vs. control, performance vs. abstraction, compile time vs. runtime, explicitness vs. conciseness. It's unclear that they will. In the long term, the reasons to create a new programming language will likely diverge significantly from the human-driven motivations of the past. There may well be an optimal programming language for LLMs—and there's no reason to assume it will resemble the ones humans have converged on. TL; DR: - Monoliths return – cheap rewriting kills dependency trees; smaller attack surface, better performance, bare-metal becomes realistic - Lindy effect weakens – legacy code loses its moat, but unknown unknowns persist; formal verification becomes essential - Strongly typed languages rise – human psychology mattered for adoption; now formal verification and RL environments favor types over ergonomics - Open source restructures – human connection drove the community; AI-written/read code breaks those incentives; alignment becomes decisive - New languages diverge – AI may not share our tradeoffs; optimal LLM programming languages may look nothing like what humans converged on ¹ x.com/mntruell/statu… ² x.com/anthropicai/st… ³ wesmckinney.com/blog/agent-erg… ⁴ #issuecomment-3717222957" target="_blank" rel="nofollow noopener">github.com/tailwindlabs/t…

Very interested in what the coming era of highly bespoke software might look like. Example from this morning - I've become a bit loosy goosy with my cardio recently so I decided to do a more srs, regimented experiment to try to lower my Resting Heart Rate from 50 -> 45, over experiment duration of 8 weeks. The primary way to do this is to aspire to a certain sum total minute goals in Zone 2 cardio and 1 HIIT/week. 1 hour later I vibe coded this super custom dashboard for this very specific experiment that shows me how I'm tracking. Claude had to reverse engineer the Woodway treadmill cloud API to pull raw data, process, filter, debug it and create a web UI frontend to track the experiment. It wasn't a fully smooth experience and I had to notice and ask to fix bugs e.g. it screwed up metric vs. imperial system units and it screwed up on the calendar matching up days to dates etc. But I still feel like the overall direction is clear: 1) There will never be (and shouldn't be) a specific app on the app store for this kind of thing. I shouldn't have to look for, download and use some kind of a "Cardio experiment tracker", when this thing is ~300 lines of code that an LLM agent will give you in seconds. The idea of an "app store" of a long tail of discrete set of apps you choose from feels somehow wrong and outdated when LLM agents can improvise the app on the spot and just for you. 2) Second, the industry has to reconfigure into a set of services of sensors and actuators with agent native ergonomics. My Woodway treadmill is a sensor - it turns physical state into digital knowledge. It shouldn't maintain some human-readable frontend and my LLM agent shouldn't have to reverse engineer it, it should be an API/CLI easily usable by my agent. I'm a little bit disappointed (and my timelines are correspondingly slower) with how slowly this progression is happening in the industry overall. 99% of products/services still don't have an AI-native CLI yet. 99% of products/services maintain .html/.css docs like I won't immediately look for how to copy paste the whole thing to my agent to get something done. They give you a list of instructions on a webpage to open this or that url and click here or there to do a thing. In 2026. What am I a computer? You do it. Or have my agent do it. So anyway today I am impressed that this random thing took 1 hour (it would have been ~10 hours 2 years ago). But what excites me more is thinking through how this really should have been 1 minute tops. What has to be in place so that it would be 1 minute? So that I could simply say "Hi can you help me track my cardio over the next 8 weeks", and after a very brief Q&A the app would be up. The AI would already have a lot personal context, it would gather the extra needed data, it would reference and search related skill libraries, and maintain all my little apps/automations. TLDR the "app store" of a set of discrete apps that you choose from is an increasingly outdated concept all by itself. The future are services of AI-native sensors & actuators orchestrated via LLM glue into highly custom, ephemeral apps. It's just not here yet.

You’ve curated the sources. You’ve researched everything. You know exactly what you want to say. You just can't get it out of your head and onto the page. We are building Almanac exactly for this. Experience the beta version here: try.almanac.so Here are a few things you can do with Almanac👇

The all-new Sarvam is here. Designed for all of India. sarvam.ai/brand

Increadible! Huge congratulations to @qualcomm_in! We would love to invest in this team to build their own company🚀 We are all in on Indian semi startups at @peakxvpartners and we would be massively excited to back this team from Qualcomm India! @SemiconIndia @AshwiniVaishnaw @GoI_MeitY

Grateful for the support of our families, friends, mentors, customers, partners, investors and the government, who believed in this vision. The journey continues!

InSoC-2 is a RISC-V chip designed and built entirely with indigenous IP at InCore.From a small team with big dreams to presenting India's homegrown processor technology to PM,this moment is a testament to whats possible when we believe in building from scratch,right here at home.

Yesterday, on National Startup Day, I had the incredible honour of interacting with the Hon'ble PM!! We showcased our Atmanirbhar semiconductor journey and presented a special memento to the PM - featuring our InSoC-2 test chip. #10YearsOfStartupIndia @startupindia @incoresemi

My yearly reminder to all founders as you go build that next Google/MSFT/Salesforce killer app in 2026: when distribution is proprietary, distribution wins (Comcast vs Netflix), when distribution is commoditized, best product wins (chrome vs IE), when product is commoditized, best service wins (Amazon vs others), when service is commoditized, best network wins.

@just_nisch_it Switch to Rust before the train gets rusty!

Another day, another null pointer error

@madmanweb @hdfcbank @Cleartrip If it's less than 24 hours, you should be entitled to a free cancellation. You can try calling the cleartrip portal number mentioned on smartbuy website.

Just had a mess of a flight booking on @HDFCBank smartbuy website, fulfilled by @cleartrip thanks to a bug in their code. Clicked "review" to go back because I had not selected all my CC points. Didn't notice the portal changed my DEL flight time from 11 AM to 0:15! (1/2)

@theBhulawat @southpkcommons This sounds cool man. Looking forward to the product!

Introducing Bunny - world's first curiosity device for kids It’s screenfree..it’s portable.. We raised $1M from @southpkcommons to reimagine how kids thrive in the age of AI, safely. Comment 'Bunny'. Our nephew will pick 50 families that get it for free this holiday season…

Public service announcement: if Salesforce hits you with steep price increases like mentioned in this Reddit, just mention the word "ZOHO" and they will drop their prices. That is their official discount code 😏 Glad to be of assistance.

To figure out who will win any given technology race, just look at the rate of acceleration of innovation & growth