Manthan Gupta
@manthanguptaa
Followers
14K
Following
24K
Media
577
Statuses
12K
building water - multi-agent orchestrator framework for fun • crafting cool things using code • creator criclang
blr/ sf
Joined March 2016
Feels legit until you see that 0.4% marker on the y-axis
You hate to see it, but 996 is real. SF-based employees are increasingly working on Saturdays as seen in corporate card spend on restaurant, takeout, and delivery. @tryramp data shows
1
0
19
Official Announcement: $UOMI is Now Live and Tradable on Uniswap and MEXC CA: 0x3628d69aa2d66e9efe95ab1267d440dec24389b6 We are excited to announce that $UOMI, the token powering the first Layer 1 built for true on-chain AI Agents, capable of owning wallets, transacting
184
319
824
Official Trailer | Forever Young- The End of Aging Has Just Begun
10
24
109
With AI adoption accelerating, we might reach a point where people build deeper connections with LLMs than with other humans, and actually prefer talking to them over real conversations.
5
0
18
The Wave of AI DefenseTech and a New Player in Rapidly Growing Market- @VWAVInc VisionWave Holdings ( Nasdaq: $VWAV) Major defense contractors like @LockheedMartin ( NYSE: $LMT) and @northropgrumman ( NYSE: $NOC) investing heavily in AI. paid content distribution
0
1
2
Folks saying they can build what Dhruv Rathee built in a day/weekend don't get it. Tech isn't the hardest part of a business, it's just an enabler. What he has: access to initial users for feedback + distribution outside the Twitter bubble = paying customers on day 1. Code is
3
2
60
Came back to blr after 3 weeks and got welcomed by heavy rain and freezing breeze
2
0
24
From Gaming to AI: Nvidia's Playbook Scales Nvidia didn’t invent a new strategy for AI—it reused an old one. This long-form piece breaks down how: - The same gaming flywheel—software first, systems next, supply control last—now powers Nvidia’s AI dominance - Hyperscalers are
0
0
5
I am planning to cross post my blogs on substack. So if you are interested in getting notified about my future blogs then you can subscribe at https://t.co/L7YbeEOOCm
3
1
23
Want better results from LLMs? It's not just "write a better prompt" Prompt engineering is steering the model in the right direction to predict the right sequence of tokens. Controlling randomness, tone, structure, and context. I broke it all down in a blog. Link in replies
3
5
64
TL;DR We want to let you in on our paid social secrets. Today we released the top 10 performance UGC ad formats of 2025 👇Comment TOP TEN for access Trust us, the top DTC brands spent $10M on our content in the last month alone
1
3
6
GPT-5 feels dumber than GPT-4o. It's like we plateaued at 4o and somehow went backwards. And now OpenAI hasdeprecated all the old models (terrible move) and I can't even attach files anymore. Peak regression.
4
0
19
Folks, buy a car if you can. Go into debt if you can because this is the state of roads in India that has recently been built. I have seen countless 2 wheeler drivers felling off their vehicle into this muddy water today. Corruption will eat this country
2
0
26
GPT 5 has a weird tendency to make functions for everything in python even when not needed
4
0
13