Top Tweets for #decentralizedknowledgegraphs
@alexutopia Hello
Probably youl find #decentralizedknowledgegraphs interesting from @origin_trail 😉
All the best in your quest !
@asyminformant @markgardn @mpcmarkets Have you had a chance to see #decentralizedknowledgegraphs work ?
Scattered information in. Verifiable agent action out.
OriginTrail gives AI agents a Shared Context Graph to act on — and every answer, transaction, and autonomous decision leaves a verifiable decision trace.
Two commands to plug in your AI agents:
🔌https://t.co/ezTB3Ic1uG
@MakisMedicine Told you they are good 😉
@umanitek powered by #decentralizedknowledgegraphs from @origin_trail !
📢 100+ fake @CredibleCrypto accounts. Taken down.
That's how this started. When CrediBULL joined the Umanitek Red Team, we detected and removed over a hundred impersonators preying on his audience.
Today, it evolves: CrediBull becomes a Umanitek product ambassador!
Expect monthly product reports in his own voice: field-testing our flagship products, and unpacking the sovereign context infrastructure they run on: the @origin_trail Decentralized Knowledge Graph.
Also ahead: a planned deep dive with Porsche-Piëch family office and Umanitek co-founders - on the vision and mission driving Umanitek in Zürich, Switzerland.
Every impersonator taken down makes the next attack harder. Every protected identity strengthens the network protecting yours.
Tune in for more!
https://t.co/HTGtFRYeat
🇺🇸 34,000+ factories. One verifiable audit layer.
Trusted Factory by @SCANAssociation, powered by Decentralized Knowledge Graph (DKG), lets major importers verify supplier security audits before goods enter the country.
Earlier trust across 40%+ of US imports.
Many bright minds have come to the same conclusion !
#AI #AIAgents #verified #memory #data #DKG #DECENTRALIZEDKNOWLEDGEGRAPHS
@origin_trail
I haven't told anyone this yet.
We are in the middle of the most consequential shift in the history of computing, and almost nobody is asking the right question.
Everyone is watching the models. The benchmarks. The context windows. The reasoning scores. I get why. Those things matter.
But they are not the frontier anymore. That moved.
The real question is not how smart a single AI can get.
It is what happens when thousands of them have to work together and trust each other.
A few weeks ago, @Karpathy pushed a 630-line Python script to GitHub and went to sleep.
He called it autoresearch.
A single AI agent. One GPU. 700 experiments in two days. It discovered 20 genuine improvements on its own, including a bug in his attention implementation he had missed after months of careful work.
The project quickly climbed to nearly 50,000 GitHub stars.
Shopify’s CEO, @tobi , ran it overnight on Shopify’s templating engine and woke up to 53% faster rendering and 61% fewer memory allocations, from 93 automated commits.
The AI world leaned in.
But Karpathy himself pointed at the real problem, and most people skimmed past it.
He described a future where agents operate as a swarm across the internet, collaborating on the same research problem, each experiment building on previous results.
Then he asked the question that stuck with me:
How do you coordinate an untrusted pool of workers on the internet?
The work is expensive to produce but cheap to verify. You have access to massive compute you cannot trust.
So how do you use it without getting poisoned by it?
He said the structure starts to resemble a blockchain. Instead of blocks, you have commits. The proof of work becomes running experiments that actually hold up.
He outlined the architecture.
He did not name the system already building toward it.
github{.}com/karpathy/autoresearch
That is what I want to talk about.
Because while the AI world debates which lab has the best memory feature, a quieter race has been running alongside it.
Most people in crypto missed it because it didn’t look like crypto.
Most people in AI missed it because it didn’t look like AI.
It looked like infrastructure.
And infrastructure rarely gets attention early.
Let me step back for a second.
The agentic AI market sits at roughly $7.6 billion today. By 2030, multiple estimates place it around $50 billion. By 2034, some projections push it to $139 billion, with growth rates above 40%.
Those numbers are not the important part.
What matters is what everything is being built on.
Right now, only about 2% of organizations have deployed agentic AI at scale. Around 60% are still exploring.
That means the infrastructure decisions being made now will define the rest of the decade.
That window is not going to stay open for long.
Every major AI company is racing to improve memory.
Anthropic added it to Claude. OpenAI integrated it into ChatGPT. Google extended it into
The demos are strong. Assistants that remember preferences, pick up where you left off, understand context over time.
Useful. Real.
But they solve a very specific loop.
One human. One assistant. One continuous thread.
The memory lives inside the platform. It creates lock-in, and the incentives behind that are obvious.
None of that leads to open coordination.
So when Agent A finishes a task and hands it to Agent B, what is actually being passed along?
If the result only exists in a session, it disappears.
If it is stored somewhere, who controls it?
Who verifies that what Agent A claims is actually what happened?
How does a third system, or a human, reconstruct the full chain of reasoning?
These are not edge cases.
They show up immediately in real systems. Financial compliance, drug discovery, medical workflows, supply chains, cybersecurity, due diligence.
The EU AI Act is already moving toward mandatory auditability.
At the same time, around 41% of code today is AI-generated, with some frontier labs approaching 90%.
Code production is accelerating.
Understanding and verifying that code is not.
Unless you can trace it.
This is the problem @origin_trail has been working on for years.
And this is where Polkadot becomes relevant, not as a narrative, but as infrastructure.
@NeuroWebAI is OriginTrail’s parachain on Polkadot.
It is where Knowledge Assets are anchored. Their fingerprints, ownership, and provenance.
Polkadot’s Elastic Scaling upgrade, live since October 2025, increased throughput across parachains while moving to flexible blockspace.
That matters here.
Because this system does not rely on trusting a platform.
It relies on verification.
Over 2 billion Knowledge Assets have already been published to the OriginTrail Decentralized Knowledge Graph.
That number was confirmed on February 25th, along with a growth curve that accelerated sharply into 2026.
These are not experiments.
Swiss Federal Railways uses it for rail safety data. Supply chains covering over 40% of US imports rely on it for compliance. Aerospace traceability across Europe. Medical verification in emerging markets. Over 200,000 auditor credentials anchored.
Grain-to-glass provenance in Ireland. Cultural heritage protection.
These are systems where accuracy matters.
They required shared memory that is not controlled by a single entity.
DKG V9 testnet shows how this extends to agents.
The team used Claude Code on a 6.8 million token monorepo, split across multiple agents.
They compared two approaches.
Agents passing notes to each other.
Agents writing and querying through the DKG.
On more complex tasks, execution was 60% faster and token cost dropped by 40%.
The gap widened as more agents were added.
Because knowledge accumulates instead of resetting.
This also addresses Karpathy’s core problem.
Agents can submit results, but those results are not accepted immediately.
They are verified.
Multiple agents reproduce and validate them. Then they are anchored cryptographically.
Only then do they enter the shared graph.
Now you have a system where results can be reused without trusting the source.
Each Edge Node in the network can act as a DKG participant, an AI agent, and a knowledge endpoint.
There are already examples of this running on lightweight hardware.
The idea is straightforward.
Computation stays local.
Verification becomes global.
V10 mainnet is expected in about four weeks based on the published roadmap.
It introduces conviction mechanisms, data monetization, and full edge participation.
The staking model rewards longer-term commitment, with higher multipliers tied to duration.
Positions are tokenized as NFTs and can be composed or transferred.
The design is deliberate.
Something the CTO, @BranaRakic , said stayed with me.
This is not competing with assistant memory.
It is a different layer.
That distinction is important.
Infrastructure develops differently.
It is slower to notice, and then difficult to replace.
I do not know exactly what gets built on top of this.
But the direction is becoming clearer.
If intelligence scales across systems, memory has to scale with it.
If decisions matter, they need to be traceable.
If systems interact, they need shared verification.
DKG V9 testnet is live.
The graph is already there.
$DOT $NEURO
DKG Node SBB on @gnosischain has reached the 5M $TRAC staking cap!
Join $TRAC token delegators to secure the Decentralized Knowledge Graph (DKG).
More AI-ready knowledge published → stronger network → greater value for all!
100 million out of 500 million is staked !
Scarcity will be noticed !
#trac #staking @origin_trail
#crypto #altcoin #gem #AI #AIAgents #RWA #DESCI #DEPIN #DKG #decentralizedknowledgegraphs #blockchain #scarcity
[NEW MILESTONE] 100+ million $TRAC staked across 3 chains: @base, @gnosischain, and @NeuroWebAI!
The @origin_trail Decentralized Knowledge Graph (DKG) keeps growing - securing knowledge and scaling global trust.
![origin_trail's tweet photo. [NEW MILESTONE] 100+ million $TRAC staked across 3 chains: @base, @gnosischain, and @NeuroWebAI!
The @origin_trail Decentralized Knowledge Graph (DKG) keeps growing - securing knowledge and scaling global trust.](https://pbs.twimg.com/media/G5ZuRETWcAE3Zm3.jpg)
The layer of #trust in #AI #data is managed on @origin_trail 's #dkg #decentralizedknowledgegraphs by @umanitek
#trac #crypto #altcoin #blockchain #gem #future #DKGcon
INSIGHTS⚡️ Guardian Agent with Umanitek detects deepfakes using DKG provenance live. Fighting AI misuse head-on seems to be the tactic! DKGcon Dec 3 teases Metcalfe's Law chapter on scaling trust, major reveals incoming? 🔍@origin_trail #TRAC #AI https://t.co/QRJazwTEk6
DKG Node umanitek 2 on @base has reached the 5M $TRAC staking cap!
Join $TRAC token delegators to secure the Decentralized Knowledge Graph (DKG)!
More AI-ready knowledge published → stronger network → greater value for all!
#trac #traceability #AI #AIAgents #RWA #DESCI #DEPIN #DKG #decentralizedknowledgegraphs #crypto #altcoin #blockchain #Uptober #gem 💎 @origintrail 👑#data
If you in $TRAC, it is getting close to breakout levels again. This is the weekly and what impresses me most about this and gives me confidence that we could break through and that downside is minimal at this point, is the volume of the last 2 week candles. Remember, there is no leverage in this token, it is all spot.
Not for @origin_trail👑#trac💎
#trac is SPOT buy ONLY traders are hedging their gains into a solid project and a stable low supply coin!
Study @origin_trail🧠
#AI #AIAgents #RWA #DESCI #DEPIN #DKG #decentralizedknowledgegraphs #crypto #altcoin #blockchain #Uptober #gem #data
That projection’s off now. Hate to say it, but it’s hard to believe we’ll see an explosive #altseason by year’s end.
This whole cycle was disaster for most #altcoins
#AIcreated #conent needs to be supervised to mitigate miss usage for ill intent!‼️
@umanitek through @origin_trail 's #dkg🧠🌐🔐 #decentralizedknowledgegraphs can be verify !📸
“We need to move fast.”
With Sora 2, deepfake and identity threats are rising fast!
@umanitek Guardian, powered by @origin_trail, lets you keep control of your data while enabling content authenticity, traceability, and stronger defenses against misuse. 🔐
The future of enterprise adoption took the stage at @PolkadotAPAC’s Arcadium, where @TomazOT & @BSI_UK’s Zhan Hao Yap shared real-world insights and approaches.
With BSI as a key partner, we’ve built several trusted solutions powered by the @origin_trail DKG! ⬇️
@BrianRoemmele The next step for LLM's are DKG's
#decentralizedknowledgegraphs
https://t.co/6h7Gfz1kJY
GenAI is powerful, but it hallucinates, inherits bias & drives up costs.
With @origin_trail Decentralized Knowledge Graph & decentralized RAG, AI agents rely on trusted, shared memories to deliver verifiable, source-based answers instead of guesses.
🧠 @BranaRakic explains!
🔍 Enter the Future of Decentralized Intelligence with #IntuitionSystems!
🌐 Your Knowledge. Your Intuition. Grow smarter as you contribute, explore, and invest. Join us in redefining Web3's potential! #Blockchain #SemanticData #DecentralizedKnowledgeGraphs
The global AI market is projected to reach $2 trillion by 2030
In other words. You dont own enough $TAO
👥🧬🔬💻⛓️✨
Innovate. Connect. Transform.
Join us September 13 at MIT Media Lab for DeSci Boston'25, uniting scientists, patients, pharma, funders, technologists, and others to rethink how science is done and shared.
August on trac(k)
🌐@umanitek & @videntifier to power the world’s largests content library on DKG
💊@TraceLabsHQ & @OxPharmaGenesis to pioneer trustworthy AI & DeSci
🎙️New On Trac(k) episode: Metcalfe Convergence Phase
Read more👇
The ticker is $BIO
Full report: https://t.co/y5vcx6Q1xF
Last Seen Hashtags on Sotwe
Most Popular Users
Elon Musk 
@elonmusk
241.5M followers
Barack Obama
@barackobama
118.8M followers
Cristiano Ronaldo
@cristiano
114.5M followers
Donald J. Trump
@realdonaldtrump
111.8M followers
Narendra Modi
@narendramodi
107.1M followers
Rihanna
@rihanna
98.7M followers
NASA
@nasa
92.3M followers
Justin Bieber
@justinbieber
91.8M followers
KATY PERRY
@katyperry
90M followers
Taylor Swift
@taylorswift13
84M followers
Lady Gaga
@ladygaga
75.5M followers
Virat Kohli
@imvkohli
73.5M followers
Kim Kardashian
@kimkardashian
70.9M followers
YouTube
@youtube
68.8M followers
Neymar Jr
@neymarjr
66.5M followers
Bill Gates
@billgates
65.3M followers
Selena Gomez
@selenagomez
63.1M followers
The Ellen Show
@theellenshow
62.2M followers
CNN
@cnn
61.7M followers
X
@x
60.7M followers