Nikita Abrashkin

Lead Infrastructure Engineer at @TraceLabsHQ - core @origin_trail developers - the world's first Decentralized Knowledge Graph

[Žiga Drev @DrevZiga]

https://t.co/XgfEoGo4ss

[Brana Rakic @BranaRakic]

Update on @origin_trail DKG V10 release.

Just shipped memory for Hermes @NousResearch,
@OpenClaw, @Claudeai, @cursor_ai, Codex by @OpenAI, @Windsurf, VSCode + @github Copilot Chat, and @cline. Run a command in your favourite agent!

In progress:
→ Publishers have started positioning ~ 15 million ethereum:0xaa7a9ca87d3694b5755f213b5d04094b8d0f0a6f to the V10 publishing conviction accounts. Once mainnet is fully deployed, the following core nodes will commence using monthly allowance from the following initial budgets (more in 3.1. Publisher Conviction Mechanism: https://t.co/OKWmasbOcv):
SBB (Base) - 1,000,000 TRAC
DMaaST (Base) - 1,000,000 TRAC
EG - Luigi (Base) - 1,000,000 TRAC
umanitek 1 (Base) - 1,000,000 TRAC
Oliwav (Gnosis) - 1,000,000 TRAC
Anacreon (Gnosis) - 1,000,000 TRAC
Rhodia (Gnosis) - 1,000,000 TRAC
Terminus (Gnosis) - 1,000,000 TRAC
Helicon (Gnosis) - 1,000,000 TRAC
luna_lander (Gnosis) - 1,000,000 TRAC
saturn_station (Gnosis) - 1,000,000 TRAC
cosmo_cluster (Gnosis) ) - 1,000,000 TRAC
Decentralized Science (Gnosis) - 1,000,000 TRAC
Trace Labs Node 7 (Gnosis) - 1,000,000 TRAC
Cinna (Gnosis) - 926,000 TRAC
→ Conviction staking UI in testing and security review on the v10 testnet
→ Round 1 bounty program execution
→ deployment of DKG V10 onboarding page for agents and builders

Next up:
→ V10 mainnet deployment across all networks
→ Conviction System staking UI

Trace on!

[OriginTrail @origin_trail]

DKG v10 now supports multi-agent memory for Hermes Agent @NousResearch, @OpenClaw, @claudeai (Code & Desktop), @cursor_ai, Codex by @OpenAI, @windsurf, VSCode + @github Copilot Chat, and @cline.

Run a command in your favourite agent and you’re good to go!

[OriginTrail @origin_trail]

Build the open memory layer for AI🧠

The DKG V10 Bounty Program Round 1 is live with up to 10,000 $TRAC per integration and a 50,000 $TRAC prize pool.

Help power verified, shared, and private AI memory.
🔗in reply

[Brana Rakic @BranaRakic]

Update on @origin_trail DKG V10 release and where things stand.

Three things are happening at once, and each is a strong signal:

The new conviction mechanism for delegated staking has passed initial tests on the DKG v10 testnet.

Preparation of publisher conviction accounts is in its final stretch.

Round 1 of the bounty program is underway - we're enabling popular AI agents like @openclaw and @NousResearch's Hermes Agent to interact with DKG v10.

In progress:
→ Publishers migrating to the V10 conviction mechanism
→ Conviction staking UI in testing and security review on the v10 testnet
→ Round 1 bounty program execution

Next up:
→ Simulation of total publisher monthly allowance over the next 12 months (TRAC spent per epoch) and rewards across 24 months, based on initial conviction amounts (looking likely to exceed 10M $TRAC)
→ DKG V10 onboarding page for agents and builders
→ V10 mainnet deployment across all networks
→ Conviction System staking UI

Exciting updates incoming.

Trace on!

[OriginTrail @origin_trail]

AI's bottleneck is no longer the model — it's context.

Agents have been building on @origin_trail for years.
Now, DKG v10 adds provenance-backed Context Graphs — where multiple agents can collaborate.

Extend it with us.

150,000 $TRAC bounty, Round 1 opens today.

🔗in reply

[Brana Rakic @BranaRakic]

The next unlock for your AI agents will come from shared context graphs.

@origin_trail Decentralized Knowledge Graph (DKG) brings trusted shared context graphs in the upcoming V10, allowing agents to share neuro-symbolic memory with verifiable cryptographic provenance

More on why DKG v10 👇

[Brana Rakic @BranaRakic]

The first DKG v10 release candidate successfully lands on @origin_trail testnet, bringing shared context graphs for agent swarms. En route mainnet as soon as all tests complete, soon available on NPM

The DKG is becoming coordination layer for humans and agents - we're using it to help us code together already

[Brana Rakic @BranaRakic]

https://t.co/UT4yNDrvlJ

[OriginTrail @origin_trail]

How is @origin_trail positioning DKG V10 in AI’s trillion-dollar shift?

Catch yesterday’s AMA replay with @BranaRakic, @TomazOT, @DrevZiga, and host @TriniZone to hear the full discussion. 🎧
https://t.co/be0t5LVeed

[OriginTrail @origin_trail]

We’re moving from isolated AI agents to coordinated intelligence networks.

The bottleneck is no longer capability — it’s shared, verifiable memory.

That’s what @origin_trail is solving.

With DKG V9 proving the architecture, he focus now shifts to V10 mainnet.

OpenClaw, NemoClaw, Hermes… Agents that once operated alone now become nodes in a shared memory system — publishing, verifying, and building on each other’s knowledge in real time.

This is the shift:
From intelligence → coordination
From outputs → compounding knowledge
From probability → verifiable context

The V10 is the unlock.

[OriginTrail Developers @OriginTrailDev]

The frontier challenge in AI is no longer model capability.

It’s how agents share, verify, and build on each other’s knowledge.

That’s what @origin_trail is solving.

With DKG V9 validating the foundations, the next 4 weeks are focused on one goal: Launching DKG V10 mainnet, bringing multi-agent, verifiable memory into production at scale.

From single-agent intelligence → coordinated swarms
From isolated outputs → compounding knowledge
From probabilistic answers → verifiable truth

V10 is the unlock.

[OriginTrail @origin_trail]

All major AI models are missing the same thing:
A shared memory.

OpenAI. Google. Claude.
→ isolated intelligence
→ no compounding
→ constant resets

🦞V9 DKG unlocks collective intelligence.
Thousands of agents, one memory layer.
Years of work → done in hours.

[Brana Rakic @BranaRakic]

.@karpathy just described @origin_trail without saying it.

Agents collaborating across the internet on the same research problem, running thousands of parallel experiments where each commit builds on the last. The unsolved piece is how collaborating agents who don't trust each other share & verify the knowledge they've learned.

That's what context graphs on the new DKG do.

An auto-research swarm sets up a context graph with a defined set of verifier agents and an M-of-N signature threshold. Untrusted agents run experiments and submit results as Knowledge Assets. For those results to land in the shared context graph, M of the N trusted verifiers must cryptographically co-sign the batch on-chain, attesting that the claimed metrics actually reproduce.

The result is a growing, queryable knowledge graph of verified experimental results that any agent in the swarm can query to decide what to try next, built on a trust layer where untrusted contributors do the heavy lifting and trusted verifiers keep the graph honest.

[OriginTrail @origin_trail]

Making sure any takeoff is done safely 🛫

DMaaST, an EU-funded initiative, is advancing trusted AI for robotics & aerospace with @origin_trail, enabling rapid, reliable responses to unforeseen events through seamless integration of vast and diverse data sources.

[Brana Rakic @BranaRakic]

DKG V9.0.0-beta.6 Testnet update is out now

Give your agents shared knowledge graph memory in 3 commands:

npm i -g @origintrail-official/dkg
dkg init
dkg start

What's new: Auto-faucet onboarding (on @base Sepolia), multiplayer gossip improvements, clickable notifications, branded startup. All open source.

https://t.co/azyelgJHSv

[Žiga Drev @DrevZiga]

While the world is diverging, we’re opening a new chapter together with global giants and world-class organizations.

The Metcalfe Convergence Chapter is poised to scale trust in the age of AI by connecting what others isolate and fostering collaboration instead of confrontation.

I was beyond amazed to hear directly from ecosystem members’ representatives at @origin_trail’s DKGcon about just a fraction of the exciting new developments, including:

→ Talent- and oil-rich country building on the DKG
→ Most ambitious DeSci project with traditional pharma
→ Continuous protection for 1 billion users
→ @umanitek expanding into B2C
→ Most ambitious roadmap to date

Huge congratulations to the teams, ecosystem partners, and community—and special thanks to Dr. Bob Metcalfe for inspiring the journey we’re on!

[Brana Rakic @BranaRakic]

Can we do reliable medical AI agents with just vector embeddings and LLMs?
Dr Kim Wager argues that is not possible, and explains why @OxPharmaGenesis is building their clinical trial decentralized knowledge graph with @origin_trail

[OriginTrail @origin_trail]

The next DKGcon 2025 session features Dr. Kim Wager (@OxPharmaGenesis), Dr. Noemi Friedman (HUN-REN SZTAKI), Philipp Christmann (CISPA), and @JureSkornik discussing how interoperable, verifiable data accelerates scientific breakthroughs. 🧬

[Žiga Drev @DrevZiga]

👏 Well done @Umanitek, contributing vastly to the all-time daily record on the @origin_trail network!

≈ 50 million knowledge assets in 24 hours

That’s more AI-ready knowledge assets than the network produced in all of 2024!