Tweet fixado
Decor Canvas - Reimagine your room with AI producthunt.com/products/launc… via @producthunt
English
Anoop
603 posts
Anoop
@AnoopCurious
Staying Curious. Exploring AI. Building https://t.co/ehAPjilXeL , https://t.co/vgoaV2RLlE , https://t.co/JC9VGUlrAf and https://t.co/5W2QJ72vW9 etc. .
Founder Land Entrou em Nisan 2025
1.2K Seguindo114 Seguidores
@Amank1412 Great! @grok could you make an equivalent list with funds and equity for startups based in India please?
English
All the best startup accelerators to apply in 2026:
1. Y Combinator ($500k for ~7%)
2. Sequoia Arc ($1M for ~10%)
3. a16z Speedrun ($500k for 10% + $500k guaranteed follow-on)
4. South Park Commons ($400k for 7% + $600k guaranteed follow-on)
5. NEO Residency ($750k uncapped SAFE, variable equity)
6. HF0 Residency ($1M uncapped SAFE for 5%)
7. Seedcamp ($350k–$1M, flexible terms)
8. Boost VC ($500k for 15%)
9. The Mint ($500k for 10%)
10. 500 Global ($150k for 6%)
11. PearX ($250k–$2M SAFE)
12. SOSV / HAX ($250k for ~7%)
13. Techstars ($220k for 5%+)
14. LAUNCH ($125k for 6%)
15. Antler US ($200k–$250k for 8–9%)
16. Founders Fellowship ($150k for 5–10%)
17. Founders, Inc. ($100k–$250k for 4–7%)
18. Antler Europe (€100k for 10% + stipend)
19. Entrepreneurs First (up to $250k for ~9%)
20. Conviction Embed ($150k uncapped MFN SAFE)
21. Afore Capital ($500k–$2M SAFE, flexible terms)
22. Soma Capital ($100k)
23. Berkeley SkyDeck ($200k)
24. gener8tor ($100k for 7.5%)
25. Heartfelt VC (up to €500k)
26. Forum Ventures ($100k for 7.5%)
27. Greylock Edge (custom SAFE + $500k+ in credits)
29. Betaworks AI Camp (up to $500k for 5% + uncapped SAFE)
29. Google for Startups (equity-free; up to $100k + $350k in cloud credits)
30. Entrepreneurs Roundtable Accelerator ($150k for 6% + $320k in credits)
31. Plug and Play Tech Center (equity-free; optional $100k–$150k SAFE)
32. Startup Wise Guys (up to €65k for equity + up to €300k follow-on)
33. HSG START Accelerator (CHF 200k for 4–10%)
34. Alchemist Accelerator (~$30k SAFE for ~5%)
35. Bethnal Green Ventures (£60,000 for 7%)
English
Insightful observations on 'world models'
Natasha Malpani 👁@natashamalpani
world models are a sexy misnomer. @ylecun , @nvidia , and @drfeifei are all building world models. they are not building the same thing. lecun is working on cognitive architecture: a system that builds causal models of reality and plans inside them. nvidia is building simulation infrastructure: physics-based environments that train, evaluate, and run physical AI systems at scale. fei-fei li is building spatial intelligence: systems that understand and reason about physical space. same term. three different bets. three different timelines. bundling them into one category inflates the hype. it also hides where value actually accumulates. the simulation moat is the most obvious. native physical interaction data was scarce. synthetic environments filled the gap. but as robotics companies accumulate real interaction data at scale, that scarcity could end. the world model framing was a data poverty artifact. the spatial intelligence bet is the most grounded. fei-fei li is trying to give machines a persistent, accurate model of physical space: where things are, how they move, what they afford. narrow, reliable, deployable. the timeline is shorter. the cognitive architecture bet is the longest. what lecun is actually building requires three components: a causal model of how the world works, a forward simulator that imagines possible futures, and a pruning mechanism. a prior over which futures are worth simulating at all. that third component does not exist in any current system. it is the difference between prediction and genuine planning. three definitions. three timelines. three completely different implications for where value accumulates.
English
Here are the notable new AI model releases from the past 7 days (approximately April 2–9, 2026), based on recent announcements:
Google Gemma 4 family (including 31B, 26B-A4B, E2B, and E4B variants) — Released around April 2 by Google DeepMind as highly capable open-source multimodal models (supporting text, image, and audio) under Apache 2.0 license, optimized for advanced reasoning, agentic workflows, and efficiency on consumer hardware.
Microsoft MAI-Transcribe-1, MAI-Voice-1, and MAI-Image-2 — Released on April 2 as in-house foundational models: MAI-Transcribe-1 is a fast, accurate speech-to-text system supporting 25 languages; MAI-Voice-1 generates high-quality custom audio voices; and MAI-Image-2 focuses on image/video generation, all made broadly available to developers.
Zhipu AI GLM-5.1 — Released around April 7 as an open-source 744B-parameter Mixture-of-Experts model excelling in coding (approaching Claude Opus 4.6 levels), long-horizon autonomous task execution (up to 8 hours), and agentic capabilities with a ~200K context window.
Meta Muse Spark — Released on April 8 by Meta Superintelligence Labs as its first frontier multimodal model from a new stack (not open weights), powering enhanced Meta AI with strong benchmarks in reasoning, multi-agent tasks, and "Contemplating Mode" for deeper problem-solving; available on meta ai with private API preview.
English
Anoop retweetou
Anoop retweetou
Was in Indore recently. And a reminder (maybe a 100th one) to VCs (incl. me) - some seriously cool startups exist outside of the Bangalore, Bombay, Delhi circuit. Here's a quick glimpse:
1. Genica Biosciences - 100% animal-free endotoxin testing via genetic engineering. Every vaccine is tested for endotoxins to avoid fatal septic shock. Under 90 min results. Zero false positives. Top 5 at BioAsia 2025.
2. Oni - AI-powered full continuum pregnancy care. 428 hospitals signed including Ujala Cygnus.
3. Aparoks - India's first prick + urine troponin detection for early cardiac emergencies. No IV needed. 1 in 4 deaths in India are heart-related.
4. MedMarvel - multimodal brain dashboard for neurologists. EEG, MRI, PET in one 3D view. 1 in 3 epilepsy patients are drug-resistant and conventional workups miss it.
5. Vyra - Ayurveda-first functional drinks for immunity, joy, focus. Going after the intentional drinking market.
6. SenseOriginal - anti-counterfeit solution for FMCG, pharma, baby care brands.
7. AutoGradeX - AI grading + student gap analysis. 2L+ users. 100+ institutions live.
8. LavaeBio - proprietary bio-leather. 95% bio-based. Himalayan fibre integration. 40-50% cheaper than Desserto. Working with Gucci, LVMH, H&M for 7+ years. Mycelium doesn't scale - this does.
Extremely bullish on what's coming out of India right now!
English
Anoop retweetou
Decor Canvas - Reimagine your room with AI producthunt.com/products/launc… via @producthunt
English
if you’re a founder raising right now
and it feels like you’re doing it alone
we built a community for you
founders sharing what’s actually working
warm intros, resources, and real connections
comment “in” and i’ll send the invite
English
The world of AI has two kinds of people
1. In the conferences and meetings all over the world
2. In the terminals cooking with Claude Code at home or their bed and some I have met on twitter/IG/ IRL are like cooking in crazy postures in 10-12 hour sprints.
The first category is arguing if AI is a hype, what will I do, this AI that AI, they are consultants, sales folks, fund managers, fund raisers, etc etc - philosophy
The others cooking are are looking at the entire surface area of all human-machine interaction - all code, all tools, all machines and just amazed at the sheer surface area of transformation and are barely trying to grasp their own learning ability.
Not kidding, I have learnt more about AI and Code and Tech in last 4 nights of cooking than I did in last 40 conferences combined.
The pace at which AI forces you to learn is SO upstream, you are literally bottlenecked by your own brains' capacity.
I am tbh so LOST in the AI universe atm, agents, skills, implementation, its a whole new way of doing everything I have ever done - talk about massive surface area, but when the dust settles, and investors or clients or users have placed their bets
1. All bets on only category 1 will give a solid 0 or likely negative of returns
2. Most bets on category 2 have high likelihood of long term sustainable value creation and returns to stakeholders.
Now the irony is ratio of 1:2 is toooo darn skewed, and with AI become cooler by the day and every Ramesh Suresh Tom Dick Chang jumping on the gun, go figure.
And the biggest googly of all - category 2 folks often lack real world understanding, business networks, the "human element", ability to sell great products, not always but often, so they gotta pair or find or partner or figure some incentive structure with category 1.
take both categories as a continuous plane, so founders/ consultants/ exec leaders will be mix of 10%*1 90%*2 - whatever plus obvious crazy imposturing - like do I know AI or do I say I know AI but I don't and I don't know I don't - hahaha go figure
So winning bets
1) Solid category 2 team with a sprinkle of category 1 - with the caveat they can work together and both sides are doing minimal imposturing maximal collaboration.
Long story short
1. AI has made building products easy but building companies and organizations - harder.
2. Investing in AI is a landmine, only those who can power through absolute immense noise and find needles in haystack will yield returns, others are well, they will be cool for the next season.
3. If you are a builder category 2, find yourself a category 1 with atleast 10-20% of category 2 skills. Basics. - So you can work together without pulling each other's hair
4. If Category 2 people can upskill and learn skills of category 1 - you da winner, you'd kill it, go get that skill.md
AI world is not gonna be as easy as everyone is making it sound. It will be a solid 15 years of buildout, maybe more. Identify your category personality, build right skills, find right partners, identify right horse to back, or right consultant for AI transformation.
Because as easy as it is to build a product, it's never been harder to build something that lasts.
English
@Its_Me_Jasraj …and run medical test results through lablingo.co.in for a free trial analysis.
English
Stop blaming genetics for hairloss. ⚠️
Before minoxidil.
Before finasteride.
Before transplant.
CHECK YOUR BLOOD. 🩸
English
@hthieblot Man! you almost got me there with your April fools post! 😃
English
Listen to me right now…
I stood right here this morning and I said I was leaving San Francisco.
I looked you all in the eye and I said… ‘I’m out. Nineteen years, finito.’
But you know what?
I’M NOT LEAVING.
I’M NOT FUCKING LEAVING!
This was gonna be my last time walking the kids to school here.
We were halfway out the goddamn door, boxes packed, kids in the car… and then I looked at this city and I said NO.
WE ARE TRIPLING DOWN!
I’m not downsizing. I’m not running. I’m moving into a giant six-bedroom triplex right here in San Francisco!
While some are cosplaying and posturing online, the hardcore builders are here, silently building the next tech for 8 billion people.
And guess what? I just leased another 25k sqft warehouse for deep tech founders!
They’re gonna need a fucking wrecking ball and the entire National Guard to evict me from my house after I stop paying the insane mortgage, go completely bankrupt and lose all my money investing in founders…
AND EVEN THEN I’M NOT LEAVING.
This is our town.
This is our moment.
And even if druggies are chasing me down the streets or anti-AI vigilantes are breaking into my house…
The game? The game is still on.
The views? I love them
The weather? Still fucking perfect.
And the show…
GOES… ON.
Because the next trillion dollar founders are all in San Francisco and nowhere else in the world…
I will find them.
I will fund them.
Let’s fucking gooooooooo!
Hubert Thieblot@hthieblot
I’m leaving SF. Everyone’s cosplaying as founders. Very few are actually building. Finito. I’ve been here for 19 years. Will miss the views and the weather. If you were starting today, where would you go?
English
