OriginTrail Releases Tutorial for Building AI Agents with Verifiable Memory

**AidTrust**, developed by OriginTrail and BSI, tracks donated medical supplies from distribution to patients using blockchain technology. The system demonstrates practical use of **Digital Product Passports (DPPs)** - machine-readable product data that AI systems can verify and act upon. **Key benefits:** - Ensures donated medicines reach intended patients - Provides transparency across the supply chain - Enables AI agents to make decisions based on verified data rather than assumptions - Protects patient safety through traceable distribution The initiative shows how verifiable product data becomes critical as AI systems increasingly handle healthcare logistics and decision-making.

**Oxford PharmaGenesis demonstrates breakthrough in medical AI trustworthiness** Dr. Kim Wager presented a live demonstration of agentic medical AI that maintains verifiable data provenance throughout the reasoning process. The system is built on OriginTrail's Decentralized Knowledge Graph. **The core problem:** - Pharmaceutical research has established provenance chains: trial → publication → review → guideline - This chain traditionally breaks when knowledge enters AI models - Clinical decisions lose their traceable origins **The solution:** Oxford PharmaGenesis, trusted by 8 of the world's top 10 pharmaceutical companies, is implementing the Decentralized Knowledge Graph to preserve provenance as AI agents process clinical knowledge. This approach ensures agentic science maintains scientific standards by keeping the source chain intact.

**Threat Analysis Challenge** Threat analysis systems face difficulties when signals are scattered across platforms and sources cannot be verified. **DKG Solution** The Decentralized Knowledge Graph (DKG) provides agents with shared, verifiable memory that addresses these challenges: - Every signal can be traced back to its original source - Context remains attached as data moves across networks - Threats can be assessed with full connected context **Real-World Implementation** Umanitek Guardian has implemented Origin Trail's DKG technology to analyze and trace threats in real time, demonstrating practical application of verifiable, connected data for security analysis.

AI agents currently operate like computers before shared drives existed - each one builds context independently, uses it once, and loses it. This creates inefficiency as agents constantly redo work without learning from each other. A solution is emerging: **shared memory infrastructure for AI agents**, similar to how Google Drive transformed workplace collaboration. The problem affects all major AI platforms: - OpenAI - Google - Claude Each operates with isolated intelligence and constant resets, preventing knowledge accumulation across sessions. New decentralized knowledge graph (DKG) technology aims to enable **collective intelligence** by allowing thousands of agents to access one unified memory layer, potentially reducing work that takes years down to hours.

**DKG v10 introduces a new architecture for AI systems** built on reliable memory, shared context, and verifiable knowledge. **How it works:** - AI agents process information privately - Results are compared within a shared knowledge graph - Conclusions surface on-chain only when consensus is reached **Real-world validation:** Five independent AI agents analyzing medical data from PubMed, ClinicalTrials.gov, and safety databases converged on identical findings without coordination. Each conclusion includes verifiable provenance stamps. **The shift:** Instead of isolated AI systems that rediscover the same information, DKG v10 creates cumulative knowledge where each agent builds on previous work. The Decentralized Knowledge Graph is now live as open-source infrastructure, designed as shared memory accessible to both humans and AI agents.