Atharva Agarwal

Building and tinkering for the new-age consumer.

Product Manager @ Meesho.
Building

[Atharva Agarwal @aatharva16]

Does anyone else feel claude ultracode is a massive token sink with little to no ROI?

[Atharva Agarwal @aatharva16]

Locked-in since last 2 months.

Checkout - https://t.co/DQJ1k1q7T5

[Atharva Agarwal @aatharva16]

Bought a 16GB Ram stick last year for <5k Rs, the same is >25k now. 5X return in a year.

[Atharva Agarwal @aatharva16]

Shifting more a 20$ to 100$ claude subscription feels the same as using Claude for the first time.

Its like a new degree of freedom!

[Atharva Agarwal @aatharva16]

@NousResearch Can this utilize existing claude code and codex subscriptions or does it only work on API?

[Atharva Agarwal @aatharva16]

@thsottiaux ChatGPT Chat becomes unresponsive way too often as the chat length increases on Desktop/Web, responses are not visible and need multiple refreshes to work.

[Atharva Agarwal @aatharva16]

@subhamCenrax @AkonLabs @ycombinator Is there a freemium model available?

[Atharva Agarwal @aatharva16]

@Joi2James MCP for Indian stock research - https://t.co/DQJ1k1q7T5

[Atharva Agarwal @aatharva16]

@pranvtwt Check out https://t.co/DQJ1k1q7T5

[Atharva Agarwal @aatharva16]

@anupamrjp Try out https://t.co/calaaX93mD

[Atharva Agarwal @aatharva16]

What's the best OCR/PDF parsing library at the moment? Looking for non-LLM based options only.

I am currently using the flow attached, any recommendations on the setup?

[Atharva Agarwal @aatharva16]

@StoxPlexShubham Checkout https://t.co/DQJ1k1q7T5 for stock research

[Atharva Agarwal @aatharva16]

Congrats on the PH launch, I am actually building something similar and saw your tool, will definitely try it out looks pretty cool.

The indian stock market data os really fragmented, and screener and other alternatives are in-sufficient for modern stock research of the AI-era. Built soemthing similar for this, would love for you to try it out here - https://t.co/NzRPyUaTxo

[Atharva Agarwal @aatharva16]

A little late to the party but I actually faced the same problem and built something similar for it - the primary interface is still built for an MCP, but encouraging to see others solve the same problem as well. Will definitely try it out.

Meanwhile you can checkout - https://t.co/NzRPyUaTxo as well

[Atharva Agarwal @aatharva16]

@deepshaswat @virattt @findatasets You can checkout https://t.co/calaaX93mD for an indian dataset

[Atharva Agarwal @aatharva16]

Today, DalalOS MCP offers:

1. 47 stock research tools
2. Coverage of approximately 5,000 NSE and BSE companies
3. Daily refreshed, timestamped data
4. No buy, sell or hold recommendations-only data and mechanically computed ratios

You ask a question in plain English. Your Al chooses the research path. DalalOS supplies the underlying numbers.

Checkout DalalOS at - https://t.co/4ngTdnCRRN

[Atharva Agarwal @aatharva16]

AI is increasingly capable of reasoning about financial markets. But when you ask it about Indian equities, it often relies on web search—surfacing whichever article won the SEO game rather than the most reliable data.

Meanwhile, information from NSE, BSE and company filings remains fragmented, difficult to access and time-consuming to reconcile.

I spent last 2-3 months building an MCP to solve the problem. DalalOS is a structured, read-only research layer built specifically for AI assistants.

You can check it out here - https://t.co/V73r658Utm

[Atharva Agarwal @aatharva16]

And well, we all know how the day ended. 🏆
Has anyone else had a moment like this on any airline?

[Atharva Agarwal @aatharva16]

Took an Air India flight yesterday that clashed with the India vs NZ World Cup Final. Everyone at the gate was glued to their phones — classic pre-flight FOMO. 🏏

[Atharva Agarwal @aatharva16]

Not sure if he was instructed to, or just couldn't help himself as a cricket fan. But that's apnapan for you — hard to explain unless you've grown up here. 🇮🇳