Mithun M

Founder and CTO -

[Mithun M @mythun]

@paraschopra Executive assistant which lives in terminal and tracks my to dos. Auto spins off agents to deliver documents for em to review based on actionable context it has. Lives in terminal

[Mithun M @mythun]

@vivekraju93 Read the play store reviews (could be growing troubles). and drivers are unhappy because of missed promises

[Mithun M @mythun]

@N01ennn Wildly token inefficient

[Mithun M @mythun]

@sangeethv_ Should use @pascalailabs

[Mithun M @mythun]

Notes from the frontier of building @pascalailabs

[Vibhav @v_vibhav]

https://t.co/PZlrgPxvyY

[Mithun M @mythun]

models are already good enough - what's required is a frontier which allows enterprises to adopt them within their constraints (interoperability, data security, ip ownership). @pascalailabs context layer makes enterprise data legible to ai for the largest funds in the world.

[Satya Nadella @satyanadella]

https://t.co/vLmiBKTtX3

[Mithun M @mythun]

Airports are where people will make you feel like youre the idiot for following rules.

[Rajneil Kamath 🚦 @rajneil]

Told a guy at the airport who broke the line, that there was indeed a line. First he pretended like there was no line. Then said I should complain to the airline. When I asked the lady at the counter why she didn’t tell him, she says “bola sir, kya karein, sunte nahi hain”. I told him to have some courtesy and some sense, so he proceeds to tell me I should go teach in a school, not him. Such folks travel abroad, and then give India a bad name!

[Mithun M @mythun]

Haven't read something this incisive in a while

[𐌁𐌉Ᏽ 𐌕𐌉𐌌𐌉 @OrevaZSN]

I saw a post on Reddit that said that “The underlying purpose of AI is to allow wealth to access skill while removing from the skilled the ability to access wealth.” And I don’t think I’ve ever seen AI described so incisively.

[Mithun M @mythun]

@aravind My god bro. This is a Wikipedia table with a citation pointing to wion news network for the source of indian road length. The wion article doesn't cite any source itself. How dumb do you think people are

[Mithun M @mythun]

If demography is destiny, what do these numbers mean? Look closely at state level differences

[Elon Musk @elonmusk]

India’s birth rate has fallen below replacement.

Among those most educated, India’s birth rate fell below replacement many years ago.

[Mithun M @mythun]

@aravind The public is already in distress. You don't know this because youre in the US.

[Mithun M @mythun]

things havent changed lol

[. @single_soul1]

Rajesh exports

[Mithun M @mythun]

Real

[Chirag @chiragiscooking]

Everyone is bullish on India until they start their own business and they need 15 papers of bhang bhosda and a current account just to integrate a payment gateway.

[Mithun M @mythun]

@RShivshankar Bigly action

[Mithun M @mythun]

@ANI Minister?

[Mithun M @mythun]

Really ridiculous stuff coming out.

[Sarthak Sidhant @sidhant_sarthak]

@ni5arga @cbseindia29 good morning CBSE, you said you used scanners to scan these copies,

now since the copies are out to the public view, do you mind explaining

which copies when scanned through a scanner, have a drop shadow? and these 3 folds?

did you really use scanners?

[www.sidin.co @sidin]

Hey I voted for culture war, why economy is not improving?????????

[Mithun M @mythun]

@deebayleaf In delhi we celebrated vishu at home, and wed invited neighbours (north Indian) for lunch. our cook had helped prepare the meal, so we asked her to dine with us. Neighbour was shocked 'shell also eat with us at the same table?'.
We thought it was normal - clearly not

[Sandeep Manudhane @sandeep_PT]

Welcome to New India.
This is straight from CJI.

[Mithun M @mythun]

The future is here - its just not evenly distributed. we see some of our most forward thinking hedge fund customers rethinking what workflows mean in the age of AI, and really generating hidden insights from their data.

[Brett Caughran @FundamentEdge]

One of my projects this year with AI in the investment process is to get this to work:
> Take a company, identify the three-key drivers that will most strongly influence the stock price
> Build agentic processes to run autonomous research on those three-key drivers, both ingesting publicly available information as well as internal research (alt data, meeting notes, idea debate transcripts, etc) that I pipe into the system
> Automatically flow insights from that 24/7/365 work back into forecasts in the model
> Agentic validation systems and click through numeric visibility (i.e. if model shows a KPI accelerating, I can click through all the source material that led to that conclusion). This creates a natural human validation check.
> This just flows back into an Investor Dashboard that the analyst & PM monitors: no exotic prompt techniques or by-hand Skills.md creation necessary. Just a column of tickers and red/green/yellow flags on the trending dimensions of these analyses
> The system outputs estimates, R/R and business momentum considerations autonomously, at scale
> Scale this to 100, then 500, then 1,000 names
> The obvious concern is that if I can do this, so can any quant firm. That's probably right. Though, like alt data, my hypothesis is there is an alpha window for this sort of alpha (which is why I'm trying to move fast, be ahead of technical capabilities, and build the Skills.md orchestration that can make these systems go when the tech is ready)
> An existential question is how much alpha will remain in primary research: in theory, the fundamental investor is getting closer to source than computerized investing ever can (i.e. a face to face conversation with management who sees weekly/monthly P&Ls and is driving initiatives to influence the KPIs), as well as scuttlebutt research with peers/suppliers/customers
> And, evaluating positioning & expectations will become ever more important, and comprehensive positioning analyses requires human support

You have to feed this with the right research & data, and as this becomes more accessible a proprietary positioning agent will be a critical adjunct to this (as in alt data, first order conclusion worked for a period, then it was all about 2nd & 3rd order considerations)

But oh man I can really see a workflow like this fundamentally changing how analyst teams spend their time.

This felt like science fiction even in fall of '25, and it's not possible today (pieces are), but with recent innovations in Agentic Workspaces (Claude Cowork, Codex and Perplexity Computer) and foundation models showing improving Excel fluency, I'm getting more bullish that workflows like this could become reality in the next 9-18 months

(if anyone building this on vendor side, or interested in building this on the fund side, please DM me)