Kamesh 🇺🇸
@ElangovanKamesh
Followers
5K
Following
6K
Media
515
Statuses
6K
Stay Open. Stay Open-Source | Angel investing in early stage | Web3 x ai Innovator | Bullish on DeSci,DeAI
San Francisco, CA
Joined October 2021
A quiet moment with Song Kang-ho. Kinda reminds me that Parasite wasn’t fiction. It was a an anatomy of invisible power, Controlled access, power without accountability. Exactly why AI must be verifiable.
3
0
2
When money is programmable without privacy, policy executes automatically. When AI operates without verifiability, governance does too.
programmable money without privacy is programmable control
8
1
14
. @SpaceX looks less like a startup success story and more like infrastructure finally catching up to ambition. The cap table just reflects who believed that early.
If SpaceX goes public at a $1T valuation, who are the biggest winners? > Elon Musk (founder) invested $100M in 2002, owns 42% = $420B > Founders Fund (Peter Thiel), invested $20M in 2008, owns 1.5% - 3% = $15B - $30B > Valor Equity Partners, invested $76M+ in 2008, owns 1.3% =
6
1
9
Instead of explaining what we’re doing, I’ll just say this: @OpenledgerHQ is building the thing every AI company skipped.
8
0
10
Futures Trader? Get 40% off + pay no activation fees. Copy trade up to 5 accounts, end-of-day drawdown in our live-market PRO+ accounts and still daily PRO payouts!
0
0
7
Seoul, here we come again 🇰🇷🐙 2025 has been unforgettable, and we’re closing the year by celebrating it with you. OpenLedger is hosting a Global-style Web3 Christmas Mixer, infused with unique Korean cultural fun. Date & Time Dec 18 (Thu) | 20:00–24:00 🎟 RSVP early. Limited
51
34
151
“Choosing” is just another word for curation under constraints. The problem isn’t intelligence anymore, it’s provenance and trust in the chooser.
AI makes knowledge cheap. That makes taste expensive. The future job is not knowing. It’s choosing.”
1
0
5
people don't hate AI they just couldn't accept the fact that they aren't as special as they thought
6
0
10
A generation of white men didn’t “fail.” They were written out. Jacob Savage’s powerful essay The Lost Generation should be required reading for anyone who still believes that Identity Politics is benign, compassionate, or cost-free... Savage documents what many young men
compactmag.com
For fifteen years I’ve scalped tickets to pay the bills. But in January 2016 I almost managed a real career.
1
1
6
Considering the current phase of AI development, I believe an AI-native university is inevitable. General-purpose models optimize for breadth, while universities optimize for depth. That gap will be filled by domain-specific models trained on verifiable, attributed data.
In 2026, I expect we’ll see the birth of the first AI-native university. Over the past several years, universities have dabbled in AI-enabled grading, tutoring, and scheduling. But what’s emerging now is deeper, an adaptive academic organism that learns and optimizes itself in
4
0
9
What replaces the current pattern is a unified system: edge inference with a coordination layer. Traditional cloud infrastructure wasn’t designed for the way agentic AI behaves. It hits bandwidth limits, creates too much data movement, and struggles when thousands of small steps
0
0
1
The Trust Factor Agentic systems require verifiable steps, not best-effort logs. Teams need to know: ▫️which model version acted? ▫️which prompt or dataset influenced a decision? ▫️whether intermediate steps were tampered with ▫️whether multi-agent coordination followed the
1
0
2
The economics - Behind The Scenes API-priced inference punishes agentic patterns. A single complex agent can trigger hundreds of model calls per task, each extending context, updating memory, and invoking external tools. Token-metered APIs turn intelligent behavior into
1
0
1
From December 22, Sonic Rumble will never be the same. Face off against Godzilla in a brand-new stage, earn Kaiju-chaotic rewards, and join the ultimate monster mayhem. 🌊 Here’s what’s lurking: - New Stage – Godzilla Survival - New Characters & Buddies - Each Kaiju brings
119
715
3K
Bandwidth becomes the first bottleneck. The reason why Agentic AI slows down is not just confined to GPUs alone. Constant data mobility that demands bandwidth also plays a crucial role. With agents in action, this becomes even heavier as they’re constantly involved in
1
0
1
The Coming Breakpoint: Agentic Systems Outgrow the Cloud Agentic AI isn’t a “bigger model.” It’s millions of small, stateful operations: vector lookups, tool calls, memory writes, planning loops, retries, and inter-agent messages. These workloads rely on tiny, rapid-fire
4
0
9
Moments like this show why verifiable AI matters. agents shouldn’t get mystery superpowers over your data. if it acts, you see the proof. trust feels real when things stay OPEN.
6
0
7
Open, composable intelligence needs open, composable data. Partnering with @ChainbaseHQ lets us extend the Hyperdata Network from passive data to active agents with proofs baked in.
We’re collaborating with @OpenledgerHQ to extend the Hyperdata Network into the AI agent layer. By combining Chainbase’s structured, AI-ready data with OpenLedger’s agent and attribution frameworks, agents can read, verify, and act with confidence in Web3. This marks another
4
0
11
🎄 Holiday server deals! Deploy selected Bare Metal Cloud servers with 30% off for the first 4 months. Use code HOLIDAY2025. Reserve for 12 months or longer & enjoy reliable, dedicated performance with cloud flexibility. Offer ends Dec 31, 2025. Don’t wait.
0
1
2
The real unlock in Decentralized AI is that it is capable of replacing the single giant centralized model paradigm with marketplaces of verifiable models, datasets, and compute. The system evolves not by bigger weights, but by composable intelligence
3
0
5
Centralized AI clusters look powerful on paper… until you realize they’re geopolitical single points of failure. The real long-term edge goes to networks that distribute compute, energy, and inference across many jurisdictions.
🐉 China’s AI edge, comes from faster build-out of power and data centers, not model tricks, argues investor Allen Zhu. “AI competition is really a competition in data centres and electricity supply and China has a significant edge there,” China is adding reactors at pace and
2
0
3
It’s funny, LLMs aren’t actually unpredictable, we just never get to see the path behind the answer. Lock that path onchain and everything gets way more sane.
Traditional LLMs feel unpredictable because nothing behind the output is stable or inspectable. In a black-box system, you can’t see what data shaped the answer, which adapter was used, or whether the model quietly changed since yesterday. OpenLedger replaces that uncertainty
4
0
6
OpenLedger isn’t expanding, we’re forking reality into new regions. New regions now run native instances: X: @OpenledgerCN TG: https://t.co/LtXfnrG8rM Onchain truth goes global!!
3
0
8
Alaska embodies the edge: vast, remote, and unforgiving. It demands technology that works where others can't. @AlaskaDOTPF's drone program reduced their critical decision-making window from 28+ hours to real time. Watch how they're reimagining what's possible with Armada at the
2
15
54