Jayashree S

@BeingPractical I feel you. One point. Not being from Ivy League - it’s still a thing. It’s prolly more of a thing now, than ever.

Old VC Pitch story from 2012 when I was building Wishberg, but I still remember this one. Two very popular super angels called me to pitch to them for a meeting at a very high-profile, super-rich bombay club. My first visit to any club of any kind whatsoever, it was intimidating anyway. I arrived early and was asked to wait until someone called for us. It was supposed to start around 7 pm or something, but I remember I had waited nearly 30-45 minutes already. Managed to glance through and these two gentlemen were sitting and already hearing the pitch of another, of a famous-celeb-founder. Nope, I am not judging - they were. Both of them were visibly drunk and laughing or cheering animatedly with that founder. They called for me. Went inside - it seemed that these angels have been there for all evening and ours seemed to be their last meeting for the day. Started the conversation, one was sloshed and was already telling his group of friends - wait just 5 mins, I will get done with this. No being from IIT or IIM or Ivy League (it was a thing back then) - had put them off already. Second one spoke about himself for 5 mins and rest of the time - cut me on every slide this won’t work, you haven’t seen the world, etc etc. 12-15 mins, and they were done. All through the meeting, had drinks, food and snacks and no one offered even glass of water or even bothered to remove food and glasses from the last pitches. The first drunk guy kept checking out passing women, and 10 mins into the pitch he left to join his friends. The second guy in all his arrogance said - buddy, do something meaningful in life. Long story short, the two gentlemen are still in the industry - over time while building Raise (knowingly or unknowingly) had tried to connect, but of course, I found a way out. Yes, the famous-founder was funded by them and over few years the venture folded silently. No big deal, we also folded Wishberg after few years. Those 15 mins of my life - I remember being treated as Bharat Bhushan in the movie Bheja Fry (in case you know, you know) who gets paraded as talent for weekend evening fun by a rich guys. I am not against strong feedback or refusing to fund - we founders face rejections all the time, but showing basic respect & humility that one should display in interactions with people is the minimum expectation.

@tejeshwi_sharma Look, I agree! The category kings of even 2030 haven’t been defined yet. But, honestly I’m tired of ppl talking about category definers and then ask you to go build for a narrow b2b requirement. What moat? How would there even be any ‘Indian tech’?

Moats are overrated in small but fast markets. A company with 40% share of a $100M market has a moat worth $40M. The same company with 10% share of a $10B market is worth 25x more and and more contestable. In India, dozens of categories will 10x in 10 years. Which means 9x of the eventual market doesn’t exist yet. The current leader is dominant over something small. The real game - the 9x - is still wide open. AI accelerates this further. A new entrant today doesn’t need to outspend the incumbent. It needs to out learn it. Foundation models are a great equaliser, the moat of “we have more engineers” just got a lot shallower. In Indian tech, the category kings of 2035 may not have been founded yet. Being first isn’t a moat. One has to earn it as the market grows.

@sonofalli A lot of things men take for granted, we have to work really hard to get. Like agency & decision making. You slog your a$$ off for years and you get hit on, she got it cuz she woman, prolly slept with the boss. You gotta be more man, than a man, to even exist here!

i love how the guy VC horror stories are like “the vc didn’t like me and tell me im special :(“ and the women are like “this guy sent me dick pics and told me to be a mother instead”

I wasn't even allowed to stay in a city where either of my parents were not there. Even when i got into good schools. I had to get into consulting & sales just to travel. Women have to work way harder to get things that are a given for men. And then this happens!

@Vineeth_Ganji @kunalb11 agreed agreed. i made this point cuz you can make bad music and walk away. but you can't make software and walk away. good or bad. the other things are huge.

@jayagotnerve @kunalb11 People mistake a functional prototype for a sustainable company, ignoring the massive operational overhead of day-to-day maintenance.

Keyboards 🎹 made it easy to make music. Didn’t make it easy to make music that people like. AI is doing the same thing to apps.

@sajithpai @davideoks @oliverwkim The China model won’t work here. I don’t want it to. It involved building a homogenous Chinese identity by essentially erasing a lot others. India is a continent masquerading as a country. Part of why India has always required its own playbook

Really good piece by @davideoks on why China has grown richer and faster than India. I do agree with his thoughts (and remember a similar piece on South Korea v Kenya by @oliverwkim). TLDR: 1/ China nuked its old social order and reorganised society around communist rule, and invested heavily in health + education. By 1980s when they liberalized its economy, it was "a socially modern country that just happened to be extremely poor". There was labour / human capital that factories could leverage, unbounded by past constraints. 2/ India on the other hand had never invested as heavily in human capital nor shifted its social order. So we havent been able to take advantage of the 1991 liberalization as much as China took advantage of Deng's easing of rules.

@adadithya Y is everyone obsessed with founders’ age? I’m in deeptech. Taking strange, bold, new bets. And idk what’s bolder than leaving a few lacs pm, zero romantic life & everything else for a conviction!

15 years in startups, & last night’s Bangerlore event was the first where I barely knew two dozen people A clear changing of guard is underway. Younger founders building in AI & deep tech, taking strange, bold, new bets Navigating people in this new order will determine success

@BrianMRey yc has some interesting stuff on this. But the best way is to live the problem. Then go talk to others who have faced the same.

how do you know if you’re building something people actually need?

@waitin4agi_ I like the work you are doing, but "just make money" is honestly..how do i put it politely.. not great advice. It's not even an option for many..

This entire “VC slept at meeting” is pointless conversation. If you want someone from someone else you have to play by those rules, even if those rules are dumb. The solution is to just be profitable and step up your ambitions over time and be patient about growth. You don’t need them.

@177pc @deedydas Maybe we all recognize the data problem & do Nation-scale data-collection drives (something like CommonCrawl-IN) - Already doing it, not open though.

I like @deedydas's work but but this take misses context Sarvam-M isn’t a vanity fine-tune; it’s India’s first open-weights 24 B Indic-centric LLM built under brutal GPU & data scarcity. Judging it by few hours of HuggingFace stats badly misses the point. Most people outside India don't appreciate that compute is quite the invisible ceiling - H100 clusters are still not commercially stocked in India - US export caps tightening next week will squeeze supply even further - Indian teams literally queue for hours of A100/H100 time that US & CN labs get on tap Data is also the long-tail problem Indic languages form <0.01 % of CommonCrawl. You read that right—two orders of magnitude less than Chinese or Spanish. Any local lab must build its corpus first, then train. That’s months of ETL before the first gradient step. Synthetic data is GPU-constrained. Talent pipeline is still forming HPC + RLHF + compiler-level optimisation is new ground in India; Sarvam’s run has already up-skilled dozens of engineers who now know how to wrangle 10 k GPU-hours, FP8 PTQ & GRPO reward engines. Their detailed blog post democratizes a lot of this learning. You can’t AWS-credit your way to that muscle memory. What Sarvam actually shipped - 3.7 M high-diversity Indic prompts, deduped & quality-scored - Two-phase non-think/think alignment that adds+2 pp on IndicGen - GRPO RL with partial-credit rewards—LiveCodeBench jump 0.23→0.44 - FP8 + look-ahead decoding: 2× tokens/s, ½ $/M tok on H100 That means a 🇮🇳-hosted midsize model now matches Gemma-3 27 B and Llama-3.3 70 B on Indic reasoning while costing a fraction to serve. That’s some engineering leverage & definitely not hype. Model adoption is anyways a long-tail - one needs to ship multiple models of non-frontier quality to eventually be able to get to the one that's truly at the frontier (at least along dimensions that we care about). Plus, there's a whole host of Indic-language use-cases where this sovereign model would work much better compared to using any other open-weights model. Look at (LiveCodeBench 0.23→0.44, 2× tokens/s) If you ask for stats, you'll learn that some of their conversational AI platform reaches out to about 50M+ people in just a week. What's next possibly? - Maybe we all recognize the data problem & do Nation-scale data-collection drives (something like CommonCrawl-IN) - Public RL-as-a-service clusters so smaller labs can replicate GRPO - For devs wanting to push the Indic NLP forward, consider forking Sarvam-M, fine-tuning on your domain corpus, benchmarking on Indic-Eval, contributing back patches. Each derivative model widens the knowledge base & closes the English–Indic gap. In summary, celebrating Sarvam's work (I'm not an investor) isn't nationalism, it's recognizing an innovation feat under constraints - India can't out-GPU Mountain View today but there's technical merit on display here, regardless of the metrics. 👏 @pratykumar, @AashaySachdeva, @HarveenChadha & other friends from @SarvamAI Here's to more AI in 🇮🇳

@nbobba I’m like 4 meetings old. Sometimes the meeting was terrible and the feedback was still useful. Sometimes the meeting was good and the feedback, not so much. When people keep asking for 1 clear icp and usecase when I say it’s deep tech, i just get a beer

In 2018, was pitching peakxv at their office. The partner badgered us over LinkedIn and email to come and meet them. We’re in a large conference room in Bangalore (has a bit of an echo), partner keeps dozing off through the meeting while the associate asks some of the most braindead questions on the planet that would likely only make sense on an IIT exam. Asked for our deck/data room even though we weren’t raising, said we weren’t raising and then asked us what we were hoping to get out of the meeting if we weren’t raising. Partner has since left, judging by track record I’m pretty confident he’s ngmi with new fund he’s trying to raise, associate is probably going to bed at night solving his IIT question bank and feeling good about himself. Same thing happened with Accel India. Not raising, associate badgers us to come by the office and meet his partner, naive founder at the time so we go. 45 min meeting on the books, partner walks in 30 min late while associate tries to hold the room politely on his behalf. No apology from the partner, couldn’t care less about us, the story or anything. We hadn’t shared a data room in advance (why would we, not raising) yells at us for wasting his time and not having any data for him to look through and proceeds to walk out of the meeting 5 min early. We politely follow up over email with thoughtful answers to some of the aggressive questions he posed during the 10 min of the meeting he attended and a thank you for his time, he never responded and neither did the associate. Partner is still at Accel India and senior leader of the firm. Good lesson for me as a VC now to know that the bar is so low at firms that the world thinks are great. Also a good reminder for us at BTV to help our companies navigate away from terrible individuals who are not worth doing business with.

@177pc I agree with this thesis. We're already seeing products built around the bottlenecks you list here. As models cross competence thresholds, opportunities shift to the systems around intelligence - data, reliability, memory.

The "Scaling Laws" paper is probably the least understood paper in AI I read this as empirical confirmation of what scaling laws have long predicted: once models reach sufficient competence at the “perspiration” layers of R&D - code synthesis, debugging, experiment execution & local iteration - the effective rate of progress becomes gated by data, compute & the remaining cognitive bottlenecks rather than raw human headcount The reported task-horizon doubling every ~4 months is consistent w/ the smooth, continuous capability curves we observed in earlier scaling work, it is not discontinuous “intelligence explosion” yet, but it is measurable compression of the outer loop Once models cross competence thresholds on execution tasks, they begin producing better data & iteration signals for themselves, compressing R&D loops in a manner fully consistent w/ smooth scaling-law extrapolation once post-training, test-time & agentic compute are included The fundamental insight & bet that Anthropic appears to have made internally is that once a model can reliably perform the constituent subtasks (code synthesis, debugging, experiment execution, local iteration), the horizon over which it can chain those subtasks without constant human intervention lengthens predictably t.co/OVVPJO7VQx

@177pc The more interesting question is why India isn't making a similar bet. Even when there is some push, founders are often encouraged to build for overseas markets from day one. 😏

The true answer is layered I think: (1) Hardware economics diverge sharply. DeepSeek V4 is explicitly optimized for & runs on Huawei Ascend 950/950PR clusters with native CANN framework support and close co-design. This bypasses (or dramatically reduces) NVIDIA’s high ASPs and ~60-75% gross margins on GPUs. US providers remain heavily NVIDIA-dependent for peak performance and ecosystem maturity; even with custom silicon progress (Google TPU, Amazon Inferentia/Trainium, Microsoft Maia), most frontier API players carry that premium. Huawei Ascend is not yet universally superior in perf/watt or absolute throughput, but the effective cost per FLOP for DeepSeek’s workloads is structurally lower - partly domestic supply chain, partly national strategic pricing (2) Architectural & systems-level inference superiority. DeepSeek’s MoE designs with extreme sparsity (reports of 27:1 activation ratios in discussion), Multi-head Latent Attention (MLA) slashing KV cache ~93%, speculative decoding, FP8, and custom redundant-expert deployment create a much lower theoretical and realized floor per token. These are not marginal tweaks; they are co-designed with the serving stack. US labs have strong caching and quantization, but fewer are pushing sparsity and KV-cache innovations at the same intensity across the board. (3) Pricing as strategic investment, not pure margin play. DeepSeek (ecosystem-aligned) appears to treat API pricing partly as customer acquisition and data/feedback flywheel fuel - classic land-grab economics. US labs balance margin signaling to investors/LPs, R&D recovery, and willingness-to-pay segmentation. Developer discourse confirms the pricing feels “too good,” raising questions of short-term cash burn vs long-term moat building It is both hardware/stack asymmetry & genuine optimization excellence, enabled by China’s full-stack sovereignty push. US players face real cost pressures and ecosystem lock-in; DeepSeek exploits the opening with talent density in systems programming and a willingness to price for adoption. This is not sustainable purely on subsidies forever, but it is durable enough to bifurcate the market.

@177pc Once a certain quality bar is achieved, that's enough for 80% of use cases. The ones who need the remaining 20% can pay more anyway. With or without subsidies, most customers care about cost economics.

@neuralunlock 😅

The bar for early stage founders is on another level right now. Easier than ever to spin up prototypes, create pitch decks, run outbound campaigns, and talk to users. Having a simple MVP is becoming table stakes. Raising with just an idea is pretty much dead.

@itsarnavb @_lagrangepoint @Venkat_lokaah

If you're NOT in Bangalore and working on the public good on any of our favorite problem areas, come to Bangerlore We at @_lagrangepoint will help with flights and accommodation, and help you meet fellow practitioners. Just DM me! - urban quality of life / civic tech - food toxicity - deep tech talent If you know someone that fits, send them my way, and I will gift you some really nice socks

@AnjneyMidha If you feel? It is moving fast! Hail scaling laws!

if you are feeling ai technology is moving too fast pls clear some room in your calendar or on weekends to learn about these tools soon in a few months, there will be a new capabilities jump (~ Sept) preparedness is key

@sethbannon If you give decent context about investment frameworks, and your domain, wouldn’t the output be fairly reasonable? Llm’s are good at reasoning if the input is good.

Founders, do not listen to this. You should in no way time your raise based on how an LLM ranks you. The best VCs are not sitting around asking AI "who should I invest in?" and then doing that.