Brana Rakic

Founder & CTO @origin_trail Decentralized Knowledge Graph | Collective Neuro-Symbolic AI | V~ℕ^2

[Brana Rakic @BranaRakic]

https://t.co/UT4yNDrvlJ

[Žiga Drev @DrevZiga]

@TheEconomist Yup, compounding value through quality context is the antidote to tokenmaxxing

[Trace Labs - Trusted Context for AI Agents @TraceLabsHQ]

Many enterprises believe they’re advancing with AI, only to hand over their institutional knowledge and learning to a few AI labs.

“The last thing any of us want is a world where every company across every sector is ceding value to a few models that eat everything they see.” @satyanadella of @Microsoft $MSFT

The real advantage isn’t just using AI. It’s owning the learning loop.

[Trace Labs - Trusted Context for AI Agents @TraceLabsHQ]

@satyanadella When companies hand over data and context to get answers back, they lose the very thing that compounds: their own evolving knowledge and judgment.

Sovereign context infrastructure is the way forward.

https://t.co/Nt6G2M53ny

[Žiga Drev @DrevZiga]

From AI memory to sovereign context infrastructure 🎯

[OriginTrail @origin_trail]

“The last thing any of us want is a world where every company across every sector is ceding value to a few models that eat everything they see.”

Sovereign context infrastructure, where enterprises own the traces and institutional memory their agents learn from, is how we avoid that outcome.

@origin_trail is building exactly that layer.

[Brana Rakic @BranaRakic]

Self sovereign technology seems like a "nice to have", until it's not.

Today it's Anthropic, tomorrow it could be any cloud service you use - good luck if all your know-how, data, APIs and services sit "somewhere" else.

Never been a better time to check out tech like @origin_trail which lets you host your context & data, so when you build on it nobody has "the switch".

Launching @origin_trail V10 next week

[Anthropic @AnthropicAI]

The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees.

The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance.

Access to all other Claude models is not affected.

We apologize for this disruption to our customers. We believe this is a misunderstanding and are working to restore access as soon as possible.

Read our full statement: https://t.co/bwn0sximKZ

[Žiga Drev @DrevZiga]

Precisely why decentralised, permissionless knowledge systems matter. @origin_trail builds them to ignore national borders and every other gatekeeper.

[OriginTrail @origin_trail]

💊The red pill for medical science: 5 AI agents, 5 siloed sources, 1 shared context graph on Decentralized Knowledge Graph.

Published literature, registered trials, real-world safety reports — different owners, different formats, no common ground.

We handed the mess to agents that never edited each other's data and asked the question that matters most in medicine: where did each claim come from, and can you check?

Working separately, they converged. On 5 disease hubs, 4 of 5 agents landed on the same condition from their own source. Every claim traceable, nothing taken on faith.

Evidence, unshackled👇

[Jurij Skornik @JureSkornik]

https://t.co/iayF4QV08M

[Žiga Drev @DrevZiga]

5 independent AI agents just converged on the same medical findings, without being told to agree. They pulled from PubMed, https://t.co/90XA84hImq, and real-world safety data, then met on a shared @origin_trail decentralized knowledge graph.

The result? Verifiable, provenance-stamped knowledge that actually holds up.

This is what trustworthy agent memory looks like.

[umanitek @umanitek]

Guardian Product Walkthrough https://t.co/xcueygECPc

[Tomaz L @TomazOT]

The @origin_trail V10 mainnet is rolling out with a bug bounty of $TRAC locked in the smart contract, ensuring the safety of the new system. Read more 👇

[dkg://TriniZone @TriniZone]

😍 OK! Let’s do it! The V10 mainnet rollout has begun.

This is a huge stepping stone into the new era of the #DKG. A network built to push the boundaries of #AI where it matters most.

The DKG is not chasing hype. It is delivering AI into mission critical environments that demand real reliability and trust. ethereum:0xaa7a9ca87d3694b5755f213b5d04094b8d0f0a6f

[OriginTrail @origin_trail]

The road to DKG V10 mainnet begins today — and it opens with the Frontier-AI Resilience Gate.

Before launch, the final V10 contracts go through an open review by independent researchers and AI-augmented teams.

Clearing this gate is the condition for going live.

Read more👇

[Jurij Skornik @JureSkornik]

AI is making attackers more capable, faster, and persistent. So the @origin_trail DKG V10 rollout starts with a public security review before mainnet.

300,000 $TRAC funded. Real honeypots. Severity rewards. Final release candidate.

This is the sequence appropriate for critical on-chain infrastructure - freeze the code, invite scrutiny, fix what is found, verify the fixes, then launch.

As AI-augmented security becomes the new baseline, the way to put it to work before mainnet, not after.

[Žiga Drev @DrevZiga]

Before @origin_trail DKG V10 hits mainnet, it goes through the Frontier-AI Resilience Gate.

Bring your models. Bring your agents. Break the final release candidate on the pre-mainnet deployment if you can. High-severity findings get rewarded.

[Trace Labs - Trusted Context for AI Agents @TraceLabsHQ]

All arguments about enterprise AI focus on which model is more capable.

It's the wrong thing to argue about.

There's a simpler question almost nobody asks out loud — and it's the one that decides everything. 👇

[Brana Rakic @BranaRakic]

Help launch the @origin_trail DKG V10 mainnet through the pre-mainnet security bounty- including a honeypot of TRAC for you to capture.

Use Fable, Opus, Codex or any other tool and try to grab it - if you manage to, it's yours!

This comes right after our latest and release candidate 17 has landed on DKG V10 testnet. More details below 👇

[OriginTrail Developers @OriginTrailDev]

The @origin_trail DKG V10 begins its mainnet rollout with a Frontier-AI Resilience Gate.

Today, the final V10 release candidate (the exact contract bytecode intended for mainnet) goes live as a public pre-mainnet, funded with 300,000 ethereum:0xaa7a9ca87d3694b5755f213b5d04094b8d0f0a6f tokens: a 200,000 TRAC honeypot pool of real, drainable positions plus a 100,000 severity-reward pool.

Independent researchers and AI-augmented teams are invited to break it: drain the honeypot and you keep what you take, and every valid finding is paid by severity. It’s a real pass/fail checkpoint: findings are fixed and verified first, and clearing the gate is the precondition for the mainnet launch. The first step of the DKG V10 deployment, by design.

Why lead with security instead of shipping and patching later?

On May 29, 2026, a researcher using @claudeai Opus 4.8 surfaced a critical, roughly four-year-old soundness flaw in @Zcash’s Orchard pool (a bug that had passed repeated expert review) in about a day, with a working proof-of-concept.

The moment matters; the trajectory matters more. Claude Mythos, Anthropic’s frontier model, is so capable at finding vulnerabilities that it was first withheld from public release and run only inside a defensive partner program, where it reportedly surfaced more than ten thousand high- and critical-severity bugs in its first month. It’s now days from a reported public release.

The bar for what an attacker, human or AI, can find only rises from here. As @AnthropicAI framed it, the advantage goes to whoever uses these tools first: attackers in the short term, defenders who fix bugs before code ships in the long term.

The Resilience Gate is how we make sure we’re on the defenders’ side, testing DKG V10 not just against today’s models but against what arrives next.

For anyone shipping on-chain systems, the implication is simple: this code launches once, mistakes can’t be undone, and the responsible move is to invite that scrutiny before any user value is at stake.

The path to mainnet, in four phases:
Phase 0: Freeze. Final contracts locked and deployed (complete)
Phase 1: Frontier-AI Resilience Gate. Open review program, through June 17
Phase 2: Mainnet launch. Hardened, feature-complete V10 (week of June 15)
Phase 3: Continuous audit. Every contract, ongoing after launch

If you work in smart-contract security, or build with AI that does, we’d welcome your review.

No allowlist, real rewards, coordinated disclosure.

*Dates are indicative: the exact mainnet date depends on the pace of network bootstrapping and the time needed to patch and re-verify any more severe findings from the Gate.

Release candidate 17 (rc17):
https://t.co/JL4nOGNGua

Bug bounty program and honeypot details:
https://t.co/rPu03hTQeo

[Žiga Drev @DrevZiga]

@PalantirTech It is also, and above all, about data enterprises tend to own versus data they hand over to AI labs.

[OriginTrail @origin_trail]

"The industry has spent a decade insisting that handing over your data is simply the cost of doing business with AI.

It never was."

Powered by DKG, @TraceLabsHQ's Network Operating System is built so that when an agent makes a decision, the answer to "who owns this" is you.