Ketan Sharma

IITK 2011 | Technophile | 3x Founder | Now coding-tinkering-designing products & producing EDM music

[Ketan Sharma @ketan_iitk]

Damn all imax 6:30 am odyssey show seats are sold out. Who is crazy waking up at 6:30 to watch Odyssey. Well tbh if there was any good seat, i would wake up or prolly sleep after watching the movie at 6:30 to 10:30 am

[Jaya Gupta @JayaGup10]

AI will disproportionately benefit ADHD minds because it externalizes the boring, parts of cognition like planning, sequencing, drafting, remembering, prioritizing and amplifies the parts ADHD minds often cook at: rapid association, novelty-seeking, pattern recognition, emotional intensity, and divergent synthesis

[Ketan Sharma @ketan_iitk]

Overthinking is biggest enemy of every competent founder

[Abhilasha Purwar @AbhilashaPurwar]

1/ In 2025 Sam Altman said India shouldn’t “waste resources” on foundational models.
Fast forward: Claude Opus 4.5 dethroned OpenAI. Proof that anyone who truly focuses can build and topple the leader.
But here’s the twist most people miss…

2/ Since then, countless users (myself included) have watched Claude’s performance collapse. My trust in Claude Code dropped 5x in the last 5 days alone. What used to be “skim & ship” is now “double-check everything.”
Add the leaked internal switch (full power only for Anthropic employees) + yesterday’s outage, and the message is clear: single-model dependency is dangerous.

3/ Meanwhile, Gemma 3 is already running locally on our machines. Not Claude-level yet, but good enough on real tasks — and 100% private. This is the real big thing: local, on-device, on-prem models that keep our data secure at both retail and enterprise level.

4/ My 5-year bet (2026–2031):The world becomes truly multimodal for consumers & enterprises.
The primary layer — memory, context graph, human thought, project knowledge — moves to local LLMs (Mac + on-prem + cloud-hosted hybrid).
Karpathy-style “thought capture” agents (currently Claude + Obsidian) will become native.

You think → local model organises → routes to the right frontier model.

The real money won’t just be in the models. It’ll be in the bridge between human and model.

5/ Last point — and this one is for India and every sovereign nation: We must invest heavily in frontier models.
Dependence on Claude (or any single foreign model) is national-security risk.
Price hikes, token cuts, model dumbing-down, or geopolitical shocks can hold entire economies hostage. Sovereign models + on-device/on-prem capability = non-negotiable.

6/ We’re doing a full circle.
From on-device → cloud → and back to on-device/on-prem.
Models sold like 1990s CDs and video games (one-time purchase) are closer than you think.

The future is multimodal, local-first, and sovereign. Who’s building the bridge? That’s where I’m putting my biggest bets.

Links
https://t.co/mX8XVqdjvm
https://t.co/QD75LDLibn
https://t.co/nirZaO6qHt

[Michael Truell @mntruell]

https://t.co/jZKOk8RAsV

[Ketan Sharma @ketan_iitk]

@rork Want!

[Ketan Sharma @ketan_iitk]

@michelleefang Yes, please! Merry Christmas, and glad to see so many people building during the holidays

[Ketan Sharma @ketan_iitk]

@calcsam Book

[Naval @naval]

Good Products are Opinionated.

“Every great founder I’ve seen up close, or even from afar, is highly opinionated and they’re almost dictatorial in how they run things.

Also, early-stage teams are opinionated. And the products they build are opinionated. Opinionated means they have a strong vision for what it should and should not do.

If you don’t have a strong vision of what it should and should not do, then you end up with a giant mess of competing features.

@Jack Dorsey has a great phrase: “Limit the number of details and make every detail perfect.” And that’s especially important in consumer products. You have to be extremely opinionated. All the best products in consumer-land get there through simplicity.

You could argue the recent success of ChatGPT and similar AI chatbots is because they’re even simpler than Google.

Google looked like the simplest product you could possibly build. It was just a box. But even that box had limitations in what you could do.

You were trained not to talk to it conversationally. You would enter keywords and you had to be careful with those keywords. You couldn’t just ask a question outright and get a sensible answer. It wouldn’t do proper synonym matching, and then it would spit you back a whole bunch of results. That was complicated. You’d have to sift through and figure out which ones were ads, which ones were real, were they sorted correctly, and then you’d have to click through and read it.

ChatGPT and the chatbot simplified that even further. You just talk to it like a human—use your voice or you type and it gives you back a straight answer.
It might not always be right, but it’s good enough, and it gives you back a straight answer in text or voice or images or whatever you prefer.

So it simplifies what we looked at as the simplest product on the Internet, which was formerly Google, and makes it even simpler. And you just cannot make a product that’s simple enough.

To be simple, you have to be extremely opinionated. You have to remove everything that doesn’t match your opinion of what the product should be doing. You have to meticulously remove every single click, every single extra button, every single setting.

In fact, things in the settings menu are an indication that you’ve abdicated your responsibility to the user. Choices for the user are an abdication of your responsibility. Maybe for legal or important reasons, you can have a few of these, but you should struggle and resist against every single choice the user has to make.

In the age of TikTok and ChatGPT, that’s more obvious than ever. People don’t want to make choices. They don’t want the cognitive load. They want you to figure out what the right defaults are and what they should be doing and looking at, and they want you to present it to them.”

[Ketan Sharma @ketan_iitk]

@ishaansehgal The best builders don’t build for VCs

[Ketan Sharma @ketan_iitk]

@natashamalpani On point indeed. Adding a couple of thoughts.

1. Conventional SaaS market is going to shrink in UI, because for repetitive workflows, AI agents are going to talk to these SaaS tools and take market share. This is where market/money can be made with AI applications.

[Ketan Sharma @ketan_iitk]

@natashamalpani .. network effects (using AI to seed one side in platforms), and also cost efficiency (AI can reduce costs to a large extent). In essence, valuations of the new age companies in AI driven market is going to be much more

[Ketan Sharma @ketan_iitk]

@natashamalpani 2. ⁠Existing business models and their moats are still going to be as relevant as always; with a new multiplier of AI automation included which can work synergistically with conventional moats like embedding (merging into user/org habits and creating high switching costs), …

[Ketan Sharma @ketan_iitk]

@PA_reinvesting True. The math is just not mathing. 50-100 people laid off makes sense but 600 is too much, what are meta AI plans ? To hire 300 of the same 600 in 6 months ?

[Kuljit @KuljitJagpal]

If anyone from the meta layoff is interested in consumer health & wellness and AI, please DM me.

We are early stage, very much bootstrapped and we are building our founding team.

[Ketan Sharma @ketan_iitk]

@evilmathkid Give it few months, quarters, and some of those no name no product we have an opening companies will be in new after raising bunch of millions.

[Rafa Felix, PhD @rafafelixphd]

Meta is hiring fast, and laying off fast.

Around 600 people were cut this week, many of them seasoned researchers and open-source contributors.

[Abhilasha Purwar @AbhilashaPurwar]

The key difference between first time founders and second time fouders is amount of time talking to VCs vs potential Customers.

As I head for Diwali catchup with friends, everyone now in 2nd 3rd 4th attempt zone is recounting last 3 months of 0$ fundraise and multiple prototyping + customer conversations.

Everyone in 1st attempt zone is VC capital raise.

Even in mails or DMs, I get from founders, 1st attempt is all idea, TAM, etc - 2nd 3rd 4th attempt is - we did x,y,z - x and y failed massively, z surprisingly picked up and grew p% yoy and this GTM channel worked best, if we put q amount of $ in this, it can likely give r% of returns.

Missing TAM, market size slides are probably very 2nd 3rd attempt features. Sure market is big, but can you capture it ? or maybe market is tiny but you can create a whole new market. The market for quick commerce was there, but not evident pre Blinkit, Zepto. TAM is probably one of the most misleading data points for considerations in early stage startup/VC. Market segments even more so. All these wonderful market segments exist, but will they convert into client segments for you, thats the key.

Client segments - now thats the real metric. Can I pick up my phone and talk to one client in each segment and hear them out. Thats gold standard of any pitch any startup can make to a VC.
Hey I am doing this, this is our financials/user growth numbers, these are details of two users/clients who are paying this much or who I can monetize in this way. I am looking to raise this much to get from point a to b.

That said, speaking and writing this easy. 5 years ago in my first attempt, even if I would have read this very post, I'd have ignored it and scrolled to next tweet 😂

[Ketan Sharma @ketan_iitk]

So glad to launch my dream of last 5-6 years - recreate lab - Venture Studio + AI transformation consulting in one roof. We work on client projects, we are CTO + tech team for founders, and we are also always building tinkering with our own products.