Sana Gupta

just a chill girl @kalaari ; also chilled @ucla

[Sana Gupta @sanantics]

most women founders I meet are SO good, but they're still out here having to work 10x harder just to get a meeting, let alone a cheque.

I've spent the last year scouting for exactly these founders. Our latest investment was in @SuperBryn_AI, an all-women founding team of total girlbosses building in the voice AI agents space. @nikkithashanker and @DrNeethuJoy are exactly who I'm talking about.

But I am not done yet and we are still looking for more such founders. honestly one of the best parts of my job is running @The_CXXO at @Kalaari, where we back women founder-CEOs ($500K–1.5M).

if you're seeing strong women-led companies raising seed/pre-seed, feel free to DM me or comment any referrals below. we move fast and we're genuinely trying to change how capital flows to women founders in India.

[Sana Gupta @sanantics]

shifting houses this week. god bless my cortisol

[Sana Gupta @sanantics]

@kalamazooooo @nova_residency Amazing job!!! 🩷🩷

[Sana Gupta @sanantics]

@rashidhamalik There’s a bunch of personal concierges that people use/ exploit like faff and kabeer (ex dunzo) is building

[Sana Gupta @sanantics]

@kalamazooooo @nova_residency Excited to work together!! This is so cool

[Sana Gupta @sanantics]

@ishandeveloper Pls no id kill for a peets mocha although the palak pathha chaat is supreme!

[Sana Gupta @sanantics]

raw cabbage on the menu today

[Sana Gupta @sanantics]

@akhil_bvs @OpenAIDevs Is there a demo day for this?? If yes would love to be a part

[Sana Gupta @sanantics]

@amritwt Facts + a great way to end it too!!

[Sana Gupta @sanantics]

@Simridhi_ Checkout the Blore local!

[Sana Gupta @sanantics]

looking for 2 bhks for rent. help. (Whitefield eek!)

[Sana Gupta @sanantics]

Romanticising the grind bc the alternative is me crashing out ;)

[Tomasz Tunguz @ttunguz]

Localmaxxing : pushing more inference to local models.

Over five weeks, I tested how much of my daily work can run on a local 35B model instead of cloud frontier models. The answer : half.

Many reasons to use local models : privacy, cost, asset depreciation.

But the only one that really matters is latency.

I ran a head-to-head benchmark. Qwen 3.6 35B-A3B-4bit on my MacBook Pro M5 vs Claude Opus 4.5 via API.

Result : 2.1x faster locally. Mean 2.8s vs 5.8s.

The local model isn't smarter. Opus scores ~20% higher on reasoning benchmarks. Local models lag frontier by 3-4 months, and for complex tasks, that gap matters.

But for routine agent tasks, it rarely does.

If half the work runs 2x faster on my laptop, I'll take that trade every time.

My little computer is about to earn its keep.

https://t.co/J2PaWEWMxs

[Sana Gupta @sanantics]

I just read meesho's full q4 letter. A lot of AI in commerce I have seen is just better recommendations on top of a product that already works. what meesho is doing is fundamentally different. PRISM serves 75% of orders through personalised feeds to users who don't know what they want to search for. Vaani lets you shop in any language without typing (1.5M users month one, 22% conversion lift.!!) GeoIndia LLM converts landmark-based addresses into coordinates because half of India's addresses don't exist in any conventional systems. Every single system is solving an access problem, not an optimisation problem. that's what makes this super interesting and can actually work to close our 970M internet users and only 30% shop online vs 80% in China gap

[Sana Gupta @sanantics]

@gautamnaray pls post the substack link to your article, looks like a great read!!

[Sana Gupta @sanantics]

@shivambhotika @WisprFlow Congrats!!! I am so happy for you Shivam!! one of the best people I have met in the VC ecosystem. Onwards and upwards!!

[Sana Gupta @sanantics]

@nikkithashanker @Kalaari Glad to be a small part of your amazing journey!

[Sana Gupta @sanantics]

@charmiekapoor This is super common in the west where people keep mini coolers outside for delivery workers! Great initiative you go girl! I think we all should follow :)

[Sana Gupta @sanantics]

@mrgirish @emergentlabs @mukundjha Thank you! It is diabolical on how people assume founders do not know how to report their own ARR?  Many companies for eg separate contracted ARR from usage/AI ARR , and use a simple, disclosed rule (e.g. trailing 3-4 months of usage × 4) to annualise variable revenue which works

[Sana Gupta @sanantics]

@shivambhotika @AcquiredFM Too cool sirji!!