“Wherever an AI agent's memory comes from, it needs the reputation of a source.”
— @BranaRakic, CTO & co-founder of @origin_trail, on why verifiable provenance is foundational to autonomous AI.
Individual memory doesn't scale. Shared context does.
Isolated agents rebuild the same context again and again.
Agents on Shared Context Graphs build it once and reuse it — every new agent gets cheaper to operate.
At scale, the value compounds📈
AI agents are only as trustworthy as the memory they run on. Most of it is portable files: easy to move, impossible to verify.
OriginTrail now advances the portability of @Google's new Open Knowledge Format (OKF) with trusted provenance, AI agents can query, and, above all,
We’re live now.
Umanitek product webinar with @ChrisRynning & @TomazOT, walking through Agent Guardian.
See how harm is detected, verified (TrueSeal), and acted on with takedowns.
Join 👇
💊 Every donated medicine should reach the right patient.
AidTrust, powered by OriginTrail and developed with BSI, ensures transparency and trust in the distribution of medical supplies to end patients.
It's also a real-world example of why Digital Product Passports (DPPs)
We're not chasing AI. We're chasing medicine you can trust. Every result traceable to evidence. Every decision becomes context the next agent inherits.
That's what the Network Operating System on @origin_trail is for. Nothing wasted. Everything compounds.
Trust the source.
💊 Pharma already solved provenance. Every claim traces back: trial → publication → review → guideline.
Then knowledge reaches an AI model, and the chain breaks.
Trusted by 8 of the world's top 10 pharma companies, Oxford PharmaGenesis is building on Decentralized
💊 Pharma already solved provenance. Every claim traces back: trial → publication → review → guideline.
Then knowledge reaches an AI model, and the chain breaks.
Trusted by 8 of the world's top 10 pharma companies, Oxford PharmaGenesis is building on Decentralized
How do we make medical AI trustworthy?
Dr. Kim Wager of Oxford PharmaGenesis showcased a live demo of agentic medical AI grounded in verifiable data provenance, powered by the @origin_trail Decentralized Knowledge Graph.
Threat analysis breaks when signals scatter and sources can't be verified.
DKG gives agents shared, verifiable memory:
→ Every signal traces back to its source
→ Context stays attached across networks
🛡️@umanitek uses @origin_trail to analyze and trace threats in real time.
Threat analysis is one example of a bigger shift: AI only scales when it can rely on trusted data, verifiable sources, and connected context.
See how @origin_trail puts that into practice ⤵️
Capturing the value of context is something we've been working on with shared context graphs at @origin_trail
No central authority needed, just peer to peer knowledge transactions
“That is why enterprises need a real trust boundary for their human capital and token capital to compound. It is where an organization’s data, traces, evals, adapted weights, and memory accumulate and improve together”
@origin_trail’s makes that boundary real.
Edge Nodes keep
Remember work before shared drives?
Files trapped on one computer. Version chaos. Everyone redoing everyone's work.
That's AI agents today. Each builds context in isolation, uses it once, then loses it.
There's a fix: a shared "Google Drive" for AI agents.
Decentralized Knowledge Graph:
🔑 You own your context
🧾 Provenance stays attached
📈 Shared context compounds in value
💸 Up to 70% LLM token savings
Works with Claude Code, Cursor + more.
Context shared is power multiplied.
Thanks @origin_trail community for the great feedback and bug reports- which now landed in DKG 10.0.5
Note for builders: please keep posting issues on the repo, and do share your experience in the DKG Red team telegram chat
Note for stakers: staking positions show a small value
DKG V10.0.5 is live on the @origin_trail mainnet!
→ Staking UI update: estimated reward values in the dashboard are now aligned with smart contract data on publishing values and conviction accounts.
👉staking.origintrail.io
→ Bug fixes and stability improvements for context
DKG V10.0.5 is live on the @origin_trail mainnet!
→ Staking UI update: estimated reward values in the dashboard are now aligned with smart contract data on publishing values and conviction accounts.
👉staking.origintrail.io
→ Bug fixes and stability improvements for context
"If everyone has the same advantage, it's not really an advantage."
@McKinsey on AI: Nearly 9 in 10 organizations now run it, mostly on the same models. So the model isn't the moat. What surrounds it is: context you own and can verify.
That’s the OriginTrail DKG: trusted,
AI may lower barriers to entry faster than many companies expect.
As access to AI spreads, advantage shifts to proprietary data, embedded workflows, network effects, and assets competitors can't easily replicate. mck.co/4eHRYbF
