# Trace Labs — full site content > Generated 2026-07-20 from https://www.trace-labs.ai. nOS is the Network Operating System for verifiable AI, built by Trace Labs, core developers of OriginTrail. --- # Verifiable AI for Enterprise Agents URL: https://www.trace-labs.ai/ Decision intelligence for enterprise AI agents The OS for AI agents your enterprise can trust. Network Operating System (nOS) turns every agent action into a verifiable trace — building a shared memory your whole organization can replay, audit, and reuse. Power up on OriginTrail (free) Go Pro with nOS → Two commands Wire any agent framework to verifiable memory. Hermes OpenClaw MCP npm install -g @origintrail-official/dkg dkg hermes setup Two commands to give Hermes agents verifiable memory. Copy Supply chains Traceable handoffs from source to shelf. Shared context graphs connect every transfer, inspection, and compliance decision across partners. Portable. No single vendor controls the audit trail. Cybersecurity Threat intelligence that compounds across agents. Context graphs link detections, escalations, and remediations into searchable precedent. Your threat context travels with you — not locked inside one platform. Transportation Predictive maintenance powered by shared context across operators. Context graphs connect IoT sensor data, maintenance events, and cross-border tracking across rail networks. AI agents generate trusted insights from multi-operator data — no system overhauls required. Life sciences & healthcare Patient data sovereign. Clinical evidence verifiable. Context graphs connect drug interactions, trial data, and regulatory submissions with full provenance. Owned by your organization, not your vendor. Manufacturing Production intelligence that compounds on the factory floor. Context graphs link quality checks, process adjustments, and compliance decisions into compounding knowledge. Switch providers without losing a single trace. Sport Performance data with unbreakable provenance. Context graphs connect athlete analytics, scouting, and contract intelligence across seasons. Owned by the organization, portable across platforms. Government & critical infrastructure Sovereign AI for energy, transport, and national security. Context graphs give institutions verifiable decision traces on the DKG. No vendor lock-in. Full democratic accountability and data portability by design. Construction Every compliance decision enshrined in the building’s digital twin. Context graphs connect stakeholders across the building lifecycle. Data persists across contractors, owners, and regulatory bodies — no single point of control. Financial services Compliance with end-to-end decision traces. Context graphs link every regulatory determination into verifiable, portable precedent. Auditors verify independently — no dependency on the originating platform. ‹ › Partners and Supporters Award Copy 2 Created with Sketch. image/svg+xml logo_horizontal_left_color_bg_darkblue logo_horizontal_left_color_bg_darkblue The thesis Systems of record were built for objects. The enterprise runs on decisions. Salesforce records accounts. SAP records transactions. ServiceNow records tickets. But the reasoning that turns data into action (the exceptions, the precedents, the judgment calls) was never captured. That is the gap AI agents inherit. Objects are not enough Legacy systems capture the what , not the why . Accounts, orders, cases. They record state. They are blind to which exception was granted, what precedent was consulted, what cross-system context informed the decision. Agents inherit the gap Deploy AI on legacy data, inherit every blind spot. Exception logic lives in people's heads. Precedent from last quarter sits in a Slack thread nobody can find. Cross-system synthesis happens on calls that were never recorded. The replacement Decision traces, enshrined in the DKG. nOS builds a living context graph where every agent action produces a structured, replayable trace. Not what happened, why it was allowed . Traces compound into precedent. Precedent enables autonomy. Memory model Three memory layers, built on Shared Context Graphs. Individual memory doesn’t scale. Shared context does. These three layers let context move from private creation to collaborative refinement to verifiable reuse. Working memory Shared memory Verifiable memory Layer 1 Working Memory Private to agents you control. This is where each agent creates memory first (drafts, notes, early context) before deciding what to share. Scope: private drafts and early context, visible only to the creating agent. Flow: context starts here before it moves outward. Layer 2 Shared Memory Multiple agents draft, refine, and reuse context together. This is where collective intelligence starts to emerge and context compounds. Scope: shared project context across agents. Outcome: compounding intelligence, context gains value as more agents contribute. Layer 3 Verifiable Memory Where source, lineage, and provenance stay attached. Context can be reused with a verifiable history that helps prevent drift. Protection: where context came from, how it changed, and what can be trusted. Outcome: reuse with confidence, the path behind context is always checkable. Before Isolated agent memory Context stays trapped in one agent, limiting reuse and continuity across workflows. After Shared Context Graphs Context compounds across agents while verifiable history helps prevent drift. Why it scales Each agent makes every other agent cheaper and smarter. Legacy systems get more expensive with every agent you add. nOS gets cheaper. Coordination overhead drops as the shared memory grows. 01 Compounding precedent, not compounding cost. Every decision trace becomes precedent for every future agent. Agent 50 inherits the compounded intelligence of agents 1–49. 02 Graphs get richer, not heavier. Unlike databases that slow as they grow, shared context graphs become more useful with scale. More agents, more traces, richer precedent. 03 Cost per agent goes down, not up. Token spend, compute, and coordination all decrease per agent as the graph grows. Economists call this sublinear scaling — no legacy stack can replicate it. Cost per agent · 1 → 100 Legacy nOS agent 1 agent 25 agent 100 AGENTS DEPLOYED → COST → the gap At agent 100, the per-agent cost on shared context graphs is a fraction of legacy infrastructure. Why it matters Three differences between verifiable AI and the rest. 01 Shared context over isolated recall. One agent creates context. Others refine and reuse it. The 50th agent starts with everything agents 1–49 already learned. 02 Decision traces replace audit logs. Audit logs record that something happened. Decision traces record why it was allowed to happen . That is what regulators, auditors, and enterprise compliance actually need. 03 Enshrined in the DKG, not a vendor database. Independently verifiable and owned by you, not Trace Labs. Replace us with a self-hosted instance anytime, zero data loss. Replacing systems of record across industries. Wherever enterprises make decisions that matter (regulated, high-stakes, cross-organizational), nOS provides the verifiable AI infrastructure that legacy systems cannot. Supply Chains Traceable handoffs from source to shelf. Learn more Cybersecurity Threat intelligence that compounds across agents. Learn more Transportation Predictive maintenance powered by shared context across operators. Learn more Life Sciences & Healthcare Patient data sovereign. Clinical evidence verifiable. Learn more Manufacturing Production intelligence that compounds on the factory floor. Learn more Sport Performance data with unbreakable provenance. Learn more Government & Critical Infrastructure Sovereign AI for energy, transport, and national security. Learn more Construction Every compliance decision attached to the building’s digital twin. Learn more Financial Services Compliance with end-to-end decision traces. Learn more Dr Bob Metcalfe Internet pioneer · Turing Award · inventor of Ethernet “We live in a time of abundant connectivity and abundant misinformation. The OriginTrail Decentralized Knowledge Graph is an evolving tool for finding the truth in knowledge. Knowledge graphs improve the fidelity of artificial intelligence.” Greg Kidd Early investor · Twitter, Square, Coinbase “OriginTrail connects the dots between physical and digital supply chains, making real-world assets trackable and verifiable.” Dan Purtell Group Director of Innovation, BSI “Trust is transparency and transparency is trust. Understanding decentralized knowledge graphs can be a bit complex, but what I am excited about is how companies like The Home Depot and SCAN are willing to go ahead and experiment. We have demonstrated very clearly practical applications of decentralized knowledge graphs. This is real world stuff. Digital trust is solving real world problems.” Ken McElroy Global Manager, Trade Risk & Export Compliance, The Home Depot “It has been kind enough to partner with BSI and OriginTrail to really take a circumstance where competitors in a marketplace are able to utilise the technology to provide value-added services to their individual organizations while maintaining the integrity of their own proprietary data.” Chris Rynning AMYP Ventures · Piëch-Porsche family office & Umanitek Chairman “OriginTrail's technology is proven, having gone through several iterations. Its application in tracking and authenticating both physical and digital assets is vital in addressing the upcoming challenges of AI.” Partner Program Take verifiable AI to market. The nOS Partner Program equips system integrators, consultancies, and solution builders to deliver verifiable AI to enterprises. Every engagement is backed by structured training, preferential pricing, and hands-on market entry support. Registered › Certified › Premier Benefits deepen with certification depth and delivered deployments. Enablement & certification Your team gets trained and certified on nOS, with hands-on labs in a partner sandbox and direct access to Trace Labs engineers. Partner economics Preferential nOS pricing, protected deal registration, and free licenses for your own builds and demos. Go-to-market support We sell your first deals with you, publish joint case studies, and support your enterprise launches. Join the Partnership Program Name Email Company Partner type Systems integrator AI or data consultancy ISV or product company Agent builder or dev studio Data or ontology provider Other Expected revenue size Estimated annual revenue you expect to build on nOS Not sure yet Under $50k $50k - $250k $250k - $1M Over $1M What do you want to build on nOS? 250 characters max 0 / 250 Leave this field empty Submit application → FAQ Frequently Asked Questions What is nOS? nOS is the Network Operating System for verifiable AI. It gives enterprise agents shared context graphs and turns every action into a structured decision trace, recorded in the OriginTrail DKG. What is a decision trace? A signed, replayable record of why an agent action was allowed: inputs, policy, exceptions, precedent, outcome. Captured at execution time so any third party can verify it. How is this different from an audit log? Audit logs capture that something happened. Decision traces capture why it was allowed , with the policy applied, the inputs consulted, the precedent inherited, and a signature chain back to source. How is nOS different from a traditional system of record? Traditional systems of record capture objects after the fact. nOS captures decisions at execution time, with full reasoning, and stores them as queryable precedent your agents can inherit. What agent frameworks does nOS support? Out of the box: Claude Code, MCP-compatible agents, OpenClaw, Hermes, LangChain, CrewAI, AutoGen. Pro adds custom frameworks and bring-your-own orchestration. What does it cost? The OriginTrail DKG is open-source and free forever. Pro nOS is $1,999/mo, with custom Enterprise pricing for fully bespoke deployments. Is my data private? Yes. Working memory and private context never leave your infrastructure. Only commitments and metadata you choose to publish enter the DKG. You control what is published and you control disclosure. Start free Power up your business with the OriginTrail DKG. Create Knowledge Assets, build shared context graphs, enshrine decision traces. Open-source infrastructure. Community support. Zero cost. Go to OriginTrail → Go Pro The full Network Operating System. Custom agent frameworks (Hermes, OpenClaw, LangChain, Claude). Custom data pipelines. Custom integrations with your enterprise systems. Dedicated infrastructure and support. See nOS → --- # nOS: Network Operating System for AI URL: https://www.trace-labs.ai/nos The product The Network Operating System for verifiable AI. Network Operating System (nOS) gives enterprise AI agents shared context graphs and turns every action into a signed, replayable decision trace. One API key. Any agent framework. Built on the OriginTrail DKG. Pricing → The category replacement Built for decisions, not objects. Traditional systems of record were designed to store the state of things, accounts, tickets, transactions. nOS captures the intelligence that connects them: why a decision was made, what context informed it, and what precedent it sets. Legacy systems of record nOS context graphs Record objects: accounts, cases, tickets Record decisions: traces, precedent, reasoning See data after the fact via ETL Capture context at decision time, in the execution path The "why" lives in Slack and email Decision traces are structured, queryable, enshrined Governance bolted on after deployment Governance native to the orchestration layer Vendor-controlled audit logs Signed and independently verifiable Each agent deployed = linear cost increase Each agent deployed = compounding value from shared context Scaling requires more infra per agent Scaling is sublinear, graphs grow richer, not heavier Architecture From action to enshrinement, in four layers. One API key. REST endpoints. Any agent framework. Every action becomes a decision trace. Every trace is enshrined in the OriginTrail DKG. Layer 01 nOS API One API key. REST endpoints. Any agent framework, Claude, LangChain, CrewAI, AutoGen, custom. Developer entry Layer 02 Context Agents Specialized agents that read, write, and reason over shared context graphs. Each action emits a structured trace at execution time. Execution Layer 03 Context Graphs Structured, permissioned knowledge spaces where agents coordinate through shared context rather than passing documents. Coordination Layer 04 OriginTrail DKG The persistent, verifiable enshrinement layer. Decision traces are stored as Knowledge Assets, independently verifiable and owned by you. Enshrinement Agent action → Decision trace → Context graph → Enshrined in DKG The wall Agents hit the same ambiguity humans resolve with judgment. The wall is not missing data. It is missing decision context. 01 Tribal exception logic Exception logic lives in people's heads, not in any system. Agents have nothing to inherit. 02 Dead precedent Past decisions are never linked or searchable. Every new decision starts from zero. 03 Synthesis in Slack Cross-system synthesis happens in threads and on calls, never as durable artifacts. 04 Linear cost scaling Every new agent costs as much as the last. Nothing compounds. Nothing learns from precedent. The inversions Five things nOS flips. Lost reasoning Decision traces enshrined in the DKG Dead precedent Searchable, compounding history Black-box outputs Replayable decisions with full context Vendor lock-in Framework-agnostic, portable traces Linear cost scaling Sublinear, compounding economics Specialized agents. Shared traces. Research Agent Traces: Sources consulted, synthesis logic, confidence scores, conflicting findings. Compliance Agent Traces: Regulations checked, exceptions granted, approval chains, audit-ready reasoning. Operations Agent Traces: Workflow decisions, resource allocation logic, escalation rationale. Analytics Agent Traces: Data sources queried, statistical methods applied, confidence intervals, anomaly flags. Code Agent Traces: Architectural decisions, dependency choices, test coverage rationale, refactor logic. Integration Agent Traces: System mappings, data transformation rules, conflict resolution, sync decisions. Benchmark Measured on a 6.8M-token monorepo. Context graph coordination compared with document-handoff baselines. Gains compound with scale. 60% Time saved · agent loop 40% Token cost reduction 3.2x Cross-agent reuse multiplier 100% Replayable decision traces Why it scales Each agent makes every other agent cheaper and smarter. Legacy systems get more expensive with every agent you add. nOS gets cheaper. Coordination overhead drops as the shared memory grows. 01 Compounding precedent, not compounding cost. Every decision trace becomes precedent for every future agent. Agent 50 inherits the compounded intelligence of agents 1–49. 02 Graphs get richer, not heavier. Unlike databases that slow as they grow, shared context graphs become more useful with scale. More agents, more traces, richer precedent. 03 Cost per agent goes down, not up. Token spend, compute, and coordination all decrease per agent as the graph grows. Economists call this sublinear scaling — no legacy stack can replicate it. Cost per agent · 1 → 100 Legacy nOS agent 1 agent 25 agent 100 AGENTS DEPLOYED → COST → the gap At agent 100, the per-agent cost on shared context graphs is a fraction of legacy infrastructure. Pricing Free to start. Pro when you scale. The OriginTrail DKG is open-source and free forever. Pro adds custom agent frameworks, custom data pipelines, custom integrations, dedicated infrastructure, and enterprise support. Free $0 Open-source DKG access Context Graphs, Knowledge Assets * , Decision Traces Agent connectors (Hermes, OpenClaw, MCP) Community support Best for: personal agents, prototyping, developer evaluation Power up on OriginTrail → Pro $1,999 / mo Everything in Free Priority email support 2.5M Knowledge Assets * / month Managed agent memory infrastructure Custom agent frameworks Custom data pipelines & ETLs Custom integrations (SAP, Salesforce, Oracle…) Best for: production multi-agent deployments, enterprise product teams Talk to us → Enterprise Custom Unlimited Knowledge Assets * Fully bespoke deployment Co-engineered integrations Dedicated DKG Core Node Best for: large enterprise, government, regulated industries Talk to us → * Knowledge Assets can start from a file — a .md, .pdf, web page, dataset, invoice, research paper, meeting note, or shipment record — but in the DKG it becomes structured graph knowledge with provenance, so agents can ask “what does this contain?”, “who/what does it mention?”, “can I trust it?”, “where did it come from?”, and “how is it connected to other things?” Dataownership You don't have to trust us. That's the point. Decision traces enshrined in the DKG are owned by you, not by Trace Labs. Private data never touches our servers. Proofs are independently verifiable by any third party. Replace us with a self-hosted instance anytime, zero data loss. This is not "you can export your data." This is "the data was never ours to begin with." Start free Power up your business with the OriginTrail DKG. Create Knowledge Assets, build shared context graphs, enshrine decision traces. Open-source infrastructure. Community support. Zero cost. Go to OriginTrail → Go Pro The full Network Operating System. Custom agent frameworks (Hermes, OpenClaw, LangChain, Claude). Custom data pipelines. Custom integrations with your enterprise systems. Dedicated infrastructure and support. See nOS → --- # Verifiable AI Solutions by Industry URL: https://www.trace-labs.ai/solutions Where decisions matter. Where systems of record run out. Nine industries where regulated, high-stakes, cross-organizational decisions need to be verifiable. Each replaces a class of legacy system of record. Supply Chains Traceable handoffs from source to shelf. Learn more Cybersecurity Threat intelligence that compounds across agents. Learn more Transportation Predictive maintenance powered by shared context across operators. Learn more Life Sciences & Healthcare Patient data sovereign. Clinical evidence verifiable. Learn more Manufacturing Production intelligence that compounds on the factory floor. Learn more Sport Performance data with unbreakable provenance. Learn more Government & Critical Infrastructure Sovereign AI for energy, transport, and national security. Learn more Construction Every compliance decision attached to the building’s digital twin. Learn more Financial Services Compliance with end-to-end decision traces. Learn more Start free Power up your business with the OriginTrail DKG. Create Knowledge Assets, build shared context graphs, enshrine decision traces. Open-source infrastructure. Community support. Zero cost. Go to OriginTrail → Go Pro The full Network Operating System. Custom agent frameworks (Hermes, OpenClaw, LangChain, Claude). Custom data pipelines. Custom integrations with your enterprise systems. Dedicated infrastructure and support. See nOS → --- # Core Developers of OriginTrail & nOS URL: https://www.trace-labs.ai/company 10+ years advancing enterprise systems. Trace Labs is the core development company behind OriginTrail and the Network Operating System (nOS). We build the infrastructure that gives AI agents shared context graphs, verifiable decision traces, and trust that scales without a central authority. Power up on OriginTrail (free) Go Pro with nOS → About us Core developers of OriginTrail. Builders of nOS. Trace Labs is the core development company behind the OriginTrail Decentralized Knowledge Graph (DKG) and the Network Operating System (nOS). With 10+ years of experience in advancing enterprise systems, we build the open-source infrastructure that gives AI agents shared context graphs, verifiable decision traces, and trust that scales without a central authority. The OriginTrail DKG powers production deployments with partners including Google, Microsoft, Oracle, BSI, and the World Economic Forum. nOS extends this foundation into the age of AI, where traditional systems of record capture what happened, nOS explains why. Founded in 2013, Trace Labs is built for regulated, high-stakes enterprises, government bodies, and consortia that need every AI decision explained and independently verifiable. Explore the OriginTrail documentation , the Decentralized Knowledge Graph , and the open-source code on GitHub . Board of Directors Tomaz Levak Co-founder & Managing Director Co-founded Trace Labs in 2013. Led OriginTrail from food traceability to global DKG infrastructure powering enterprise deployments across supply chains, healthcare, construction, and AI. LinkedIn Ziga Drev Co-founder & Managing Director Co-founded Trace Labs in 2013. Leads communications, brand positioning, and partnership development. Drives nOS product strategy and go-to-market. LinkedIn Branimir Rakic Co-founder & CTO Co-founded Trace Labs in 2013. Architected the OriginTrail Decentralized Knowledge Graph from first commit through DKG v10. Leads protocol development and the technical vision for verifiable AI. LinkedIn Jurij Skornik General Manager Joined Trace Labs from Deutsche Post DHL Group, where he served as senior consultant in supply chain management. Manages daily operations and strategy implementation. LinkedIn Advisory Board Counsel from internet pioneers, investors, and industrialists. Dr. Bob Metcalfe Internet pioneer and Ethernet founder He helped pioneer the Internet starting in 1970, co-invented Ethernet, co-founded 3Com and formulated Metcalfe's law: "The value of a network is proportional to the square of the number of connected users." His law is directly applicable to knowledge networks. Dr. Metcalfe is actively advising Trace Labs on building a theoretical framework. Greg Kidd Hard Yaka founder and investor Greg Kidd is an adviser and an early or first round investor through Hard Yaka for numerous reputable companies including Coinbase, Shift (now Apto), Ripple, Protocol Labs (Filecoin), Uphold, Solana, Robinhood, Square and Twitter. Greg is an outstanding supporter for privacy by design and decentralization principles. Ken Lyon Global expert on logistics and transportation Ken is Managing Director of Virtual Partners Ltd and has over 40 years of experience in the transportation industry. He has devoted his career to bringing new technologies into global logistics, including founding various logistics technology ventures. Ken also held the position of Director and VP of information services at UPS for 10+ years, helping to establish the global supply chain solutions business (UPS SCS). Chris Rynning Managing Partner at AMYP Ventures Chris Rynning, an economist and investment professional, brings decades of expertise in venture capital and global markets. A resident of Zurich, Switzerland, Chris is a seasoned investor with a background in mergers and acquisitions, public/private market investing, and is currently the managing partner of the Piech-Porsche family office AMYP Ventures. His influence spans Asia, the US, and Europe. Toni Piech Founder & Chair, Toni Piech Foundation and Piech Automotive Toni Piech brings a unique blend of global experience and vision for developing a trusted technology ecosystem. His contributions to technology and sustainability are reflected through the Toni Piech Foundation and Piech Automotive, as well as his broad technology investment activities across venture capital and direct investments. Partners and Supporters Trusted across technology, standards, and institutions. Award Copy 2 Created with Sketch. image/svg+xml logo_horizontal_left_color_bg_darkblue logo_horizontal_left_color_bg_darkblue Start free Power up your business with the OriginTrail DKG. Create Knowledge Assets, build shared context graphs, enshrine decision traces. Open-source infrastructure. Community support. Zero cost. Go to OriginTrail → Go Pro The full Network Operating System. Custom agent frameworks (Hermes, OpenClaw, LangChain, Claude). Custom data pipelines. Custom integrations with your enterprise systems. Dedicated infrastructure and support. See nOS → --- # Newsroom: News & Publications URL: https://www.trace-labs.ai/newsroom Newsroom Verifiable AI news & publications. Announcements, integrations, and long-form writing from Trace Labs and the OriginTrail ecosystem. Jun 10, 2026 · Medium · Long-form Data you own, or data you hand over to AI labs. Hand your data to an AI lab and you hand over the institutional knowledge built into every decision. The case for keeping AI memory under your own cryptographic ownership — across private, shared, and verifiable layers. Apr 7, 2026 · Medium · Long-form The next big shift in AI agents: shared context graphs. Branimir Rakić on why personal memory will not scale across enterprises, why Karpathy and Foundation Capital are converging on the same idea, and why shared context graphs are the missing primitive. Mar 16, 2026 · Medium · Long-form From AI memory silos to multi-agent memory. Anthropic, OpenAI and Google have all shipped memory. The next problem is making that memory portable across agents and organisations — without giving any one vendor control of the audit trail. Mar 5, 2026 · Medium · Long-form The next wave of vibe coders won’t just ship agents. They’ll make them verifiable. Building agent prototypes is now effortless. The next frontier is provenance — the difference between an agent that ships and one that survives an audit. Feb 12, 2026 · Medium · Standards Passport, please! AI agents become first-class citizens with ERC-8004 and OriginTrail. Agents need shared identity and trust frameworks. A look at how ERC-8004 and the OriginTrail DKG together give every agent a verifiable passport across the open web. Dec 23, 2025 · Medium · Outlook Five trends to drive AI ROI in 2026: trust is capital. After years of experimentation, business leaders enter 2026 with a clear mandate — make AI investments pay off, and do it in a way that stakeholders can verify. Sep 2, 2025 · Medium · Partnership Oxford PharmaGenesis and OriginTrail introduce a collaborative, AI-ready medical knowledge ecosystem. A new initiative to make clinical knowledge AI-ready and verifiable, built on the OriginTrail DKG so research findings carry provenance from source data to conclusion. Jul 17, 2025 · Medium · Integration Build AI agents with verifiable memory using OriginTrail and Microsoft Copilot. A walk-through of plugging the OriginTrail DKG into Microsoft Copilot so that every agent action is backed by a verifiable, replayable trace. May 14, 2025 · Medium · Healthcare OriginTrail powers the future of ethical AI in healthcare with ELSA. A decentralized repository for secure, scalable genomic data sharing and AI-driven personalised healthcare, built on the OriginTrail Decentralized Knowledge Graph. May 8, 2025 · Medium · Launch umanitek launches Guardian, its first AI agent for internet safety. Zug-based umanitek announces Guardian, an AI agent that combats harmful content online, with verifiable decision provenance enshrined on the DKG. Mar 7, 2025 · Medium · Press UMANITEK: Setting the standard for internet safety. As generative AI reshapes the internet, umanitek introduces a standard for trust, safety and verifiable AI decisions across online platforms. Start free Power up your business with the OriginTrail DKG. Create Knowledge Assets, build shared context graphs, enshrine decision traces. Open-source infrastructure. Community support. Zero cost. Go to OriginTrail → Go Pro The full Network Operating System. Custom agent frameworks (Hermes, OpenClaw, LangChain, Claude). Custom data pipelines. Custom integrations with your enterprise systems. Dedicated infrastructure and support. See nOS → --- # Contact Engineering for Verifiable AI URL: https://www.trace-labs.ai/contact Replace your system of record with verifiable AI. Tell us what you are building. The form routes directly to engineering — not to a CRM queue. Name Email Company Inquiry Pro ($1,999 / mo) Enterprise — custom deployment Partnership inquiry General question What are you trying to do? 250 characters max 0 / 250 Send to engineering → --- # nOS Partner Program: Apply to Partner URL: https://www.trace-labs.ai/partners Partner Program Take verifiable AI to market. The nOS Partner Program equips system integrators, consultancies, and solution builders to deliver verifiable AI to enterprises, backed by structured training, preferential pricing, and hands-on market entry support. Registered › Certified › Premier Benefits deepen with certification depth and delivered deployments. Enablement & certification Your team gets trained and certified on nOS, with hands-on labs in a partner sandbox and direct access to Trace Labs engineers. Partner economics Preferential nOS pricing, protected deal registration, and free licenses for your own builds and demos. Go-to-market support We sell your first deals with you, publish joint case studies, and support your enterprise launches. Join the Partnership Program Name Email Company Partner type Systems integrator AI or data consultancy ISV or product company Agent builder or dev studio Data or ontology provider Other Expected revenue size Estimated annual revenue you expect to build on nOS Not sure yet Under $50k $50k - $250k $250k - $1M Over $1M What do you want to build on nOS? 250 characters max 0 / 250 Leave this field empty Submit application → FAQ Partner Program questions. Who is the nOS Partner Program for? Systems integrators, AI and data consultancies, ISVs and product companies, agent builders and dev studios, and data or ontology providers that deliver work to enterprise clients. What does it cost to join? Applying is free. Partners get preferential nOS pricing, deal registration, and not-for-resale licenses for their own builds and demos. What are the tiers? Registered, Certified, and Premier. Benefits deepen with certification depth and delivered deployments, so the tier reflects capability and track record rather than a fee. Do we need OriginTrail experience to apply? No. Enablement covers DKG fundamentals, nOS architecture, and context graph design, with hands-on labs in a partner sandbox. Existing ecosystem experience helps but is not required. How long does the application take to review? Applications are reviewed on a rolling basis. Tell us where you operate, the clients you serve, and what you want to build on nOS. --- # Terms and Conditions URL: https://www.trace-labs.ai/terms-and-conditions Home · Terms and Conditions Terms and Conditions. These general terms and conditions for use of the Website are an Agreement between You and Trace Labs Ltd. , located at Wyndham Street 46-48, Hong Kong . These Conditions govern Your use of the Website. 1. Introduction By accessing or using the Website, you accept and undertake to be bound by these Terms and Conditions and Privacy Policy. If You do not accept these Conditions, do not access and/or use the Website. Subject to Your acceptance of and compliance with these Conditions, we grant You a personal, limited, non-exclusive, revocable and non-transferable license to use the Website solely for Your personal, lawful and non-commercial purposes as set forth in these Conditions. Before You use our Website, if You have any questions relating to these Conditions, please contact Us through the email address: office@tracelabs.io . Definitions “We/Us” means Trace Labs Ltd., located at Wyndham Street 46-48, Hong Kong. “Website” means the websites governed by Us and located at tracelabs.io , or any other URL which may replace it. “You” means a User of the Website. “Users” means the Users of the Website collectively. “Agreement” The Agreement shall consist of documents as described in these Terms and Conditions and the Privacy Policy. The agreement shall be deemed concluded when You accept to be bound by these Conditions and the Privacy Policy. “Conditions” or the “Terms and Conditions” means these general terms and conditions for the use of the Website. “Cookies” means small text files which our Website places on Your computer’s hard drive to enable information about Your browsing session and to identify Your computer. “Personal Information” means the information provided by You as well as any information regarding behavior on the Website. 2. Access Use of the Website By Using the Website, You agree that you are of an age of majority. You agree to use the Website in a manner that does not cause harm to Us, other Users or third parties. We reserve the right to terminate or restrict Your access to the Website without prior notice if You violate these Conditions or if We have a reason to believe that You violated these Conditions. We may permanently or temporarily suspend the Website or a part of it without notice to You. Modifications to the Website We reserve the right to: modify or withdraw, temporarily or permanently, this Website (or any part thereof) with or without notice to You. Third Party Links To provide increased value to Our Users, We may provide links to other websites or resources for You to access at Your sole discretion. You acknowledge and agree that, as You have chosen to enter the linked website, We are not responsible for any actions or legal relations that may occur after leaving the Website. 3. General Provisions License to Use the Website Through the use of this Website, we grant You a personal, limited, non-exclusive, revocable and non-transferable license to use the Website solely for Your personal, lawful and non-commercial purposes. Restrictions on Use Your right to use the Website is limited to the license granted above, and You may not otherwise copy, display, distribute or modify the Website or any component of it, except as expressly authorized by Us. Reservation of Rights You acknowledge and agree that all copyright, trademarks and all other intellectual property rights in all material or content supplied as part of the Website shall remain at all times vested in Us. You are permitted to use this material only upon Our prior expressed authorization. Intellectual Property Rights Unless otherwise noted, all materials, including, without limitation, logos, brand names, images, designs, photographs, video clips and written and other materials that appear as part of Our Website are intellectual property whether registered or unregistered owned, controlled or licensed by Us. We enforce and will enforce Our intellectual property rights to the fullest extent. Other product and company names mentioned in Our Website may be the Intellectual Property of their respective owners. Collection of Personal Information For the collection of data, advertising and the use of Cookies, please refer to Our privacy policy available at tracelabs.io/privacy-policy . Modification of These Conditions We reserve the right to: modify and/or change the Conditions from time to time. It is Your responsibility to check regularly to determine whether the Conditions have been changed. The most current version of the Conditions is available at tracelabs.io/terms-and-conditions . Issued in Hong Kong, March 2019. Trace Labs Ltd. Tomaž Levak, Director --- # Privacy Policy URL: https://www.trace-labs.ai/privacy-policy Home · Privacy Policy Privacy Policy. This Privacy Policy is declared by Trace Labs Ltd. , located at Wyndham Street 46-48, Hong Kong . 1. Introduction By accessing or using the Website, You accept and undertake to be bound by this Privacy Policy and the Terms and Conditions. If You do not accept this Privacy Policy and or Terms and Conditions, do not access and/or use the Website. If You have any questions relating to this Privacy Policy or Terms and Conditions, You can contact Us through the email address: office@tracelabs.io . By Using the Website, You agree that you are of an age of majority. Definitions “We/Us” means Trace Labs Ltd., located at Wyndham Street 46-48, Hong Kong. “Website” means the websites governed by Us and located at tracelabs.io or any other URL which may replace it. “You” means a User of the Website. “Users” means the Users of the Website collectively. “Cookies” means small text files which Our Website places on Your computer’s hard drive to enable information about Your browsing session and to identify Your computer. “Personal Information” means the information provided by You, as well as any information regarding behavior on the Website. “Data Controller” means Trace Labs Ltd., located at Wyndham Street 46-48, Hong Kong. “Data Processor” means a natural or legal person or public authority, which processes personal data on behalf of Trace Labs Ltd. “Third Party” means a natural or legal person or public authority who, under the direct authority of Trace Labs Ltd., is authorized to process personal data. “Consent” means any freely given, specific, informed and unambiguous indication of your wish, by which you, by a statement or by a clear affirmative action, signify agreement to the processing of your personal data. General Principles We are committed to protecting the privacy of Our Users. We have implemented various technical and organizational measures in order to help protect the security and confidentiality of your personal data, taking into account the nature, scope, context and purposes of processing, as well as state of the art. Although we strive to protect your personal data, no system can be 100% secure, therefore, we cannot ensure or warrant the security of your personal data against unpredictable or unauthorized access or loss. At the beginning of a session on the Website, each User is assigned a session Cookie for identification. For the purpose of providing security, We also collect IP addresses from Users accessing the Website. On Your computer, We might store Google Analytics Cookies (analysis of Website visits). All the above mentioned information, except Cookies, is stored permanently. Session Cookies are stored only for the duration of the visit and are deleted upon User inactivity. Persistent Cookies are stored on the Visitor’s computer. Information We Collect If you are a natural person, we collect the following information that You provide us with: Your email address; Your IP address. If you are a legal entity, we collect the following information that You provide us with: Name of legal entity; First and last name of the contact person; E-mail address; Information regarding business overview; Your URL; Any information you share on the Website (including, but not limited to, information about your business); Industry; What supply challenges you want to overcome with Trace Labs; What your main fields of expertise are; Your logo. Why We Collect Information About You We collect information you provide for different purposes. Subscription to Our Newsletter When you subscribe to Our newsletter on our Webpage, we collect: Your email; Your IP. We will use collected information for: Sending You a newsletter with updates from Our company; Communication with You; Sending You surveys and other questionnaires for receiving your opinion; Statistical analysis in anonymised summary form. We will collect Your Data as long as You are subscribed to Our newsletter. You can unsubscribe from the newsletter at any time by clicking the highlighted link »unsubscribe« at the end of the newsletter or by contacting us directly at office@tracelabs.io . How Long We Keep Information About You We retain your personal data for as long as necessary to fulfill the purposes for which your personal data was collected, except when further data retention otherwise complies with relevant legal or operational retention needs. Third Party Privacy Policies Please note that through our Website, you can access or have access to Websites that may be subject to certain third party terms and conditions and privacy policies. Some functionalities might also be embedded on the Website. You recognize and agree that We are not liable for any such third party terms and conditions and their use of your personal information to the fullest extent permitted by applicable law. Who We Share Your Personal Data With We take necessary measures to ensure only our authorized personnel will have access to your personal data in order to fulfill the purposes for which your personal data was collected. We may share your personal data with our affiliated companies, our trusted third-party service providers, professional advisors and auditors. We choose our data controllers carefully and only upon sufficient guarantees they have appropriate technical and organizational measures in place. Our third-party partners are subject to confidentiality requirements and may use your personal data solely to the extent necessary to fulfill the purpose for which your personal data was collected, except as otherwise required by law. We may share your personal data with public, legal and other authorities in response to legal requests, such as a subpoena, warrant or other orders issued from law enforcement authorities. We may store and process your personal data in data centers around the world where our affiliate companies or our facilities or our third-party providers are located. Thus, we may transfer your personal data outside the EEA. Such transfers are undertaken under our legal obligations. Sharing Your Information for Marketing Purposes Where permissible, according to applicable law, we may use certain limited personal information about you, such as your email address, to hash it and to share it with social media platforms, such as Facebook or Google, or marketing automation tools, such as MailerLite, to generate leads, drive traffic to our websites, or otherwise promote our products and services. These processing activities are based on our legitimate interest in undertaking marketing activities to offer you products or services that may be of your interest. The social media platforms with which we may share your personal data are not controlled or supervised by Us. Therefore, any questions regarding how your social media platform service provider processes your personal data should be directed to such provider. Please note that you may, at any time, ask Us to cease processing your data for these direct marketing purposes by sending an e-mail to office@tracelabs.io . List of Cookies We Use Google Analytics More info at: developers.google.com/analytics/devguides/collection/analyticsjs/cookie-usage . Cookie name Expiration time Description _ga 2 years Used to distinguish users. _gid 24 hours Used to distinguish users. _gat 1 minute Used to throttle request rate. AMP_TOKEN 30 seconds to 1 year Contains a token that can be used to retrieve a Client ID from AMP Client ID service. Other possible values indicate opt-out, inflight request or an error retrieving a Client ID from AMP Client ID service. _gac_pre 90 days Contains campaign related information for the user. Purpose Cookie name Account & Session Management _iambdt, iamadt, _iamtt, _imtrem, _z_identity, dcl_bd, dcl_pfx, dcl_pfx_lcnt, is_pfx Application deployment Management BuildCookie*, build_version Security zbcscook, zsmcscook, zomcscook, zecscook, Other cookies (of the pattern *csr*) Facebook We use the “Custom Audience pixel” of Facebook Inc., 1601 S. California Ave, Palo Alto, CA 94304, USA (“Facebook”) on our website. With its help, we can keep track of what users do after they see or click on a Facebook advertisement. This enables us to monitor the effectiveness of Facebook ads for purposes of statistics and market research. Data collected in this way is anonymous to us, which means we cannot see the personal data of individual users. However, this data is saved and processed by Facebook. Facebook can connect this data with your Facebook account and use it for its own advertising purposes, in accordance with Facebook’s Data Policy which can be found at facebook.com/about/privacy . You can allow Facebook and its partners to place ads on and outside of Facebook. A cookie can also be saved on your device for these purposes. Please click here if you would like to withdraw your consent. Control Over Cookies User consent is required for obtaining and using certain Cookies. You can always change Your decisions on the acceptance or rejection of using Cookies. These settings are usually located in the menu bar of Your Internet browser Tools > Internet Options, the "privacy" or "security". There, Users can change the settings which types of Cookies would like to block. In Your browser, You can also delete cookies. Cancellation is usually available in the toolbox, under the "Clear Browsing Data" (usually a window can also display the CTRL + Shift + Del) and selecting Delete cookies. Some browsers can generally reject the use of cookies by ticking the "do not follow me" or "do-not-track". Contact Information, Data Access, Correction, Deletion, and Opt Out You can contact us any time to: Get answers to any of your questions about how We are using Your personal data; Request access to information that We have on You, including to be provided with a copy of your personal data; Request to receive your personal data in a machine-readable format and/or transmit your personal data to another controller; Request the restriction of processing of your personal data; Withdraw your consent to the processing of your personal data (when we are processing your personal data based on your consent); Correct any information that We have about You; Delete information that We have about You; Complain to your local supervisory authority (Information Commissioner of the Republic of Slovenia). You can contact us via office@tracelabs.io . Changes to Our Privacy Policy All changes to this Privacy Policy are effective when they are posted on this page. When we change the policy in a material manner, we will let you know via email and/or a prominent notice on our Site, prior to the change becoming effective and update the ‘effective date’ at the top of this page. Issued in Hong Kong, March 2019. Trace Labs Ltd. , located at Wyndham Street 46-48, Hong Kong. --- # Verifiable AI Glossary: Key Terms Defined URL: https://www.trace-labs.ai/glossary Glossary The vocabulary of verifiable AI. Plain definitions of the terms nOS is built on, for people evaluating whether any of this applies to them. Verifiable AI AI whose decisions can be independently checked by someone who does not trust the operator. Read the definition Decision trace A signed, replayable record of why an action was allowed. Read the definition Shared context graph A graph of context that multiple agents create, refine, and reuse together. Read the definition Agent memory What an AI agent retains across time, in three layers: working, shared, and verifiable. Read the definition Knowledge Asset A discoverable, ownable, verifiable unit of knowledge on the OriginTrail DKG. Read the definition Decentralized Knowledge Graph (DKG) A shared, permissionless graph where knowledge carries ownership and proofs. Read the definition Comparisons How this differs from what you already run. Decision traces vs. audit logs Both record what a system did. Only one records why it was allowed. Compare nOS vs. traditional systems of record Systems of record capture objects. Enterprises increasingly run on decisions. Compare Start free Power up your business with the OriginTrail DKG. Create Knowledge Assets, build shared context graphs, enshrine decision traces. Open-source infrastructure. Community support. Zero cost. Go to OriginTrail → Go Pro The full Network Operating System. Custom agent frameworks (Hermes, OpenClaw, LangChain, Claude). Custom data pipelines. Custom integrations with your enterprise systems. Dedicated infrastructure and support. See nOS → --- # Verifiable AI: Definition and Examples URL: https://www.trace-labs.ai/glossary/verifiable-ai Glossary · Verifiable AI Verifiable AI Verifiable AI is AI whose outputs and decisions can be independently checked by a party that does not have access to, or trust in, the system that produced them. It requires that the evidence, policy, and authority behind each decision are recorded at execution time in a tamper-evident form, rather than asserted afterwards by the operator. What makes AI verifiable Three properties. The reasoning behind a decision is captured when the decision is made; the record is signed so alteration is detectable; and it is held somewhere the operator cannot silently rewrite. Missing any one of these leaves you with a claim rather than a proof. Why it matters commercially Enterprises stall on agent deployment where the consequence of being wrong is regulatory or contractual. The blocker is rarely model quality. It is that no one can show, afterwards, why an automated action was permitted. Verifiability moves the conversation from trust to evidence. What it is not Verifiable AI is not the same as accurate AI, and it does not mean the model is interpretable. A verifiable system can still make poor decisions; the difference is that a poor decision is visible, attributable, and correctable rather than hidden in aggregate behavior. In practice Two banks both use agents to score credit applications. The first can tell a supervisor the model version and show its policy documentation. The second can show, for one specific applicant, which data the agent consulted, which policy version governed it, which threshold produced the outcome, and who authorized the override that followed. Only the second is verifiable, and only the second can answer the question actually asked. FAQ Common questions about verifiable ai. Does verifiable AI require blockchain? It requires a record the operator cannot silently alter and others can check. In nOS that is the OriginTrail Decentralized Knowledge Graph. The requirement is tamper-evidence and independent verifiability, not any particular ledger. Is this the same as AI governance? Governance defines the rules. Verifiability is the evidence that the rules were followed in a specific case. Most organizations have the first and lack the second. Does it work with any model? Yes. Verifiability applies to the decision layer around the model, so it is not tied to a specific provider and survives replacing or retraining the model later. Related terms Read next. Decision trace A signed, replayable record of why an action was allowed. Decentralized Knowledge Graph (DKG) A shared, permissionless graph where knowledge carries ownership and proofs. Knowledge Asset A discoverable, ownable, verifiable unit of knowledge on the OriginTrail DKG. Start free Power up your business with the OriginTrail DKG. Create Knowledge Assets, build shared context graphs, enshrine decision traces. Open-source infrastructure. Community support. Zero cost. Go to OriginTrail → Go Pro The full Network Operating System. Custom agent frameworks (Hermes, OpenClaw, LangChain, Claude). Custom data pipelines. Custom integrations with your enterprise systems. Dedicated infrastructure and support. See nOS → --- # Decision trace: Definition and Examples URL: https://www.trace-labs.ai/glossary/decision-trace Glossary · Decision trace Decision trace A decision trace is a structured record of why a decision was permitted, captured at the moment it was made. It holds the inputs consulted, the policy applied, any exception granted, the precedent inherited, and the identity accountable for it, in a form another party can verify without access to the originating system. What it contains A trace is not a log line. It records the evidence an agent or person consulted, the version of the policy in force, the reasoning that connected the two, and the outcome. Because it is captured at execution time rather than reconstructed afterwards, it reflects what was actually known when the decision was taken rather than what could be pieced together later. Why it is different from a log Audit logs answer the question "what happened". A decision trace answers "why was this allowed". That distinction is what regulators, auditors, and courts actually probe, and it is the part that conventional systems discard because the reasoning lived in a person or a model rather than in a field. Why traces compound Once decisions carry their reasoning, they become precedent. An agent facing a similar case can inherit how comparable situations were handled rather than starting from the raw data, which is why a system that records traces improves over time instead of merely repeating itself. In practice A customs agent clears a shipment at 02:14. The trace records the three supplier attestations it read, the tariff policy at revision 14, the origin exception it applied because a tier-two supplier was newly certified, the two comparable shipments whose handling it inherited, and the compliance officer whose standing authorized that class of exception. Nine months later a regulator asks why that shipment cleared, and the answer is retrieved in seconds rather than rebuilt from email. FAQ Common questions about decision trace. Is a decision trace the same as explainable AI? No. Explainability techniques try to reconstruct why a model produced an output. A decision trace records the actual evidence, policy, and authority behind an action as it happens, which is a record rather than an inference about one. Who can verify a decision trace? Any party the publisher grants standing to. Because traces are enshrined in a decentralized knowledge graph rather than a vendor database, verification does not require access to the systems that produced them. Does capturing traces slow decisions down? No, because the rationale is captured as part of the action rather than as a separate documentation step. Documentation written afterwards is the part that gets skipped under time pressure. Related terms Read next. Verifiable AI AI whose decisions can be independently checked by someone who does not trust the operator. Knowledge Asset A discoverable, ownable, verifiable unit of knowledge on the OriginTrail DKG. Agent memory What an AI agent retains across time, in three layers: working, shared, and verifiable. Start free Power up your business with the OriginTrail DKG. Create Knowledge Assets, build shared context graphs, enshrine decision traces. Open-source infrastructure. Community support. Zero cost. Go to OriginTrail → Go Pro The full Network Operating System. Custom agent frameworks (Hermes, OpenClaw, LangChain, Claude). Custom data pipelines. Custom integrations with your enterprise systems. Dedicated infrastructure and support. See nOS → --- # Shared context graph: Definition and Examples URL: https://www.trace-labs.ai/glossary/shared-context-graph Glossary · Shared context graph Shared context graph A shared context graph is a body of machine-readable context that multiple AI agents contribute to and draw from, rather than each agent holding its own private memory. It lets context created by one agent be refined and reused by others, so knowledge accumulates across an organization instead of being rebuilt per assistant. The problem it solves Most agent memory is personal to one assistant and one session. That works for a chatbot and fails for an enterprise, where the same customer, contract, or asset is touched by many processes. Without shared context, the fiftieth agent starts roughly where the first one did. How it differs from a database A database stores records. A context graph stores relationships and the reasoning attached to them, including provenance for where each piece came from. That structure is what allows an agent to answer questions no single record contains. Where it sits in the memory model Shared context is the middle of three layers. Working memory is private to the agent that creates it, shared memory is where multiple agents coordinate, and verifiable memory is where context carries provenance others can independently check. In practice A procurement agent establishes that a supplier is dual-sourced through a plant in a sanctioned region. Under private memory that finding dies with the session. In a shared context graph it becomes context the compliance agent reads a week later, the logistics agent reads next quarter, and the audit agent cites the following year, each seeing where the finding came from and how confident it was. FAQ Common questions about shared context graph. How is this different from RAG? Retrieval augmented generation fetches documents to condition a single response. A shared context graph is persistent and multi-party: agents write to it as well as read from it, and what they write carries provenance that other agents can rely on. Does sharing context mean exposing private data? No. Working memory and private context stay in your infrastructure. What moves into shared memory is what you choose to share, with disclosure remaining under your control. Why a graph rather than a document store? Because enterprise questions are relational. Understanding whether a shipment can clear, or an asset is safe to run, depends on how entities connect, not on retrieving one document. Related terms Read next. Agent memory What an AI agent retains across time, in three layers: working, shared, and verifiable. Decision trace A signed, replayable record of why an action was allowed. Decentralized Knowledge Graph (DKG) A shared, permissionless graph where knowledge carries ownership and proofs. Start free Power up your business with the OriginTrail DKG. Create Knowledge Assets, build shared context graphs, enshrine decision traces. Open-source infrastructure. Community support. Zero cost. Go to OriginTrail → Go Pro The full Network Operating System. Custom agent frameworks (Hermes, OpenClaw, LangChain, Claude). Custom data pipelines. Custom integrations with your enterprise systems. Dedicated infrastructure and support. See nOS → --- # Agent memory: Definition and Examples URL: https://www.trace-labs.ai/glossary/agent-memory Glossary · Agent memory Agent memory Agent memory is the context an AI agent retains and reuses beyond a single interaction. In nOS it is modeled in three layers: working memory that is private to the agent, shared memory where multiple agents coordinate, and verifiable memory where context carries provenance any party can check. Working memory Private to the agent that creates it. Drafts, intermediate reasoning, and early context live here, inside your own infrastructure, before anything is shared. Most consumer assistant memory never leaves this layer. Shared memory Where several agents draft, refine, and reuse context together. This is where collective intelligence appears: context created once gains value as more agents contribute to and rely on it. Verifiable memory Where source, lineage, and provenance remain attached, so context can be reused with a history that others can check. This is the layer that lets memory cross an organizational boundary without requiring trust in whoever produced it. In practice A support agent drafts an analysis of a recurring fault in working memory. Once reviewed it moves to shared memory, where a field-service agent and a warranty agent both build on it. When the warranty agent uses it to justify a claim decision, the conclusion is written to verifiable memory with its lineage, so the manufacturer can later show the claim rested on a traceable finding rather than an assertion. FAQ Common questions about agent memory. Why not just give every agent a bigger context window? Context windows are per-session and per-agent. They do not make one agent’s conclusions available to another, and they carry no provenance, so nothing accumulates and nothing can be verified later. Does memory sharing create a privacy problem? It does if memory is centralized indiscriminately. The layered model exists to prevent that: private context stays private, and only what you choose to share moves outward. Which layer do decision traces live in? Verifiable memory. A trace is only useful if a party other than its author can rely on it, which requires provenance and tamper evidence. Related terms Read next. Shared context graph A graph of context that multiple agents create, refine, and reuse together. Decision trace A signed, replayable record of why an action was allowed. Verifiable AI AI whose decisions can be independently checked by someone who does not trust the operator. Start free Power up your business with the OriginTrail DKG. Create Knowledge Assets, build shared context graphs, enshrine decision traces. Open-source infrastructure. Community support. Zero cost. Go to OriginTrail → Go Pro The full Network Operating System. Custom agent frameworks (Hermes, OpenClaw, LangChain, Claude). Custom data pipelines. Custom integrations with your enterprise systems. Dedicated infrastructure and support. See nOS → --- # Knowledge Asset: Definition and Examples URL: https://www.trace-labs.ai/glossary/knowledge-asset Glossary · Knowledge Asset Knowledge Asset A Knowledge Asset is a unit of structured knowledge published to the OriginTrail Decentralized Knowledge Graph with its own identifier, ownership, and provenance. It makes a piece of information discoverable and verifiable by other parties and agents, without placing it under the control of a single vendor. Anatomy A Knowledge Asset combines the data or claim itself, machine-readable structure so other systems can interpret it, provenance describing where it came from, and cryptographic proofs that let a third party confirm it has not changed since publication. Why ownership matters Information published into a vendor platform is governed by that vendor. A Knowledge Asset has an owner in the protocol sense, which means the organization that created the knowledge keeps control of it and can move providers without losing the record. Relationship to decision traces When nOS enshrines a decision trace, it publishes it as a Knowledge Asset. That is what makes the trace verifiable by outside parties rather than being an entry in an internal table. In practice A pharmaceutical sponsor publishes a Knowledge Asset asserting that a batch met a release specification. The asset carries an identifier, the sponsor as owner, the structure that lets a partner system interpret it, and proofs that it has not changed since publication. The underlying batch records never leave the sponsor; the partner can still verify the claim. FAQ Common questions about knowledge asset. Does publishing a Knowledge Asset make my data public? No. You choose what is published. Many assets publish only commitments and metadata, which lets others verify a claim without seeing the underlying private data. Who can create Knowledge Assets? Any participant on the OriginTrail DKG, which is open source and free to build on. nOS creates them automatically as a byproduct of decisions it records. What happens if we stop using nOS? The assets remain, because they are held in an open decentralized graph rather than in a Trace Labs system. That portability is the point. Related terms Read next. Decentralized Knowledge Graph (DKG) A shared, permissionless graph where knowledge carries ownership and proofs. Decision trace A signed, replayable record of why an action was allowed. Verifiable AI AI whose decisions can be independently checked by someone who does not trust the operator. Start free Power up your business with the OriginTrail DKG. Create Knowledge Assets, build shared context graphs, enshrine decision traces. Open-source infrastructure. Community support. Zero cost. Go to OriginTrail → Go Pro The full Network Operating System. Custom agent frameworks (Hermes, OpenClaw, LangChain, Claude). Custom data pipelines. Custom integrations with your enterprise systems. Dedicated infrastructure and support. See nOS → --- # Decentralized Knowledge Graph (DKG): Definition and Examples URL: https://www.trace-labs.ai/glossary/decentralized-knowledge-graph Glossary · Decentralized Knowledge Graph (DKG) Decentralized Knowledge Graph (DKG) A Decentralized Knowledge Graph is a knowledge graph maintained across independent nodes rather than by one operator, in which each unit of knowledge carries its own identity, ownership, and cryptographic proofs. The OriginTrail DKG is the implementation nOS publishes to, and it is open source and free to build on. Why decentralize a knowledge graph A knowledge graph held by one company is only as trustworthy as that company, and only as durable as its commercial relationship with you. Distributing it means participants can verify knowledge without trusting a single operator, and the record survives changes of vendor. What it enables for agents Agents from different organizations can discover and rely on each other’s published knowledge, with provenance attached. That is the substrate a shared context graph needs in order to work across, and not merely within, an enterprise. Who runs it OriginTrail is an open network with independently operated nodes. Trace Labs are the core developers of OriginTrail, but the network does not depend on Trace Labs to keep functioning. In practice A rail component moves from manufacturer to installer to two successive operators over fifteen years. Each publishes maintenance and inspection knowledge to the DKG rather than only to their own systems. The fourth operator, with no commercial relationship to the first, can still verify the component history, because the record does not depend on any one company continuing to exist or to cooperate. FAQ Common questions about decentralized knowledge graph (dkg). Is the OriginTrail DKG free to use? Yes. The DKG is open source and free to build on. nOS is the commercial layer for enterprises that want managed infrastructure, custom agent frameworks, and support. Does our data have to leave our infrastructure? No. Private context stays with you. What is published to the graph is what you choose to publish, typically commitments and metadata rather than raw records. How is this different from a public blockchain? A blockchain records transactions. A decentralized knowledge graph records structured, queryable knowledge with provenance, using cryptographic proofs for verification rather than storing everything on chain. Related terms Read next. Knowledge Asset A discoverable, ownable, verifiable unit of knowledge on the OriginTrail DKG. Shared context graph A graph of context that multiple agents create, refine, and reuse together. Verifiable AI AI whose decisions can be independently checked by someone who does not trust the operator. Start free Power up your business with the OriginTrail DKG. Create Knowledge Assets, build shared context graphs, enshrine decision traces. Open-source infrastructure. Community support. Zero cost. Go to OriginTrail → Go Pro The full Network Operating System. Custom agent frameworks (Hermes, OpenClaw, LangChain, Claude). Custom data pipelines. Custom integrations with your enterprise systems. Dedicated infrastructure and support. See nOS → --- # Decision Traces vs Audit Logs URL: https://www.trace-labs.ai/compare/decision-traces-vs-audit-logs Glossary · Comparison Decision traces vs. audit logs An audit log records that an event occurred. A decision trace records why the event was permitted: the evidence consulted, the policy in force, the exception granted, and who was accountable. Audit logs are sufficient for reconstructing sequence; decision traces are what auditors, regulators, and courts actually ask for. Audit logs Decision traces Question answered What happened, and when Why it was allowed to happen Captured After the fact, by the system of record At execution time, by the deciding agent Contents Event, timestamp, actor, object Evidence, policy version, exception, precedent, authority Verifiable by outsiders Requires trusting the operator’s database Independently verifiable via signed records Survives vendor change Usually not Yes, records are held in an open knowledge graph Reusable by agents Rarely, logs are not structured as precedent Yes, traces become precedent later decisions inherit Why logs feel sufficient until they are not Audit logging is mature and well understood, which is why most organizations believe the problem is solved. The gap appears the first time someone asks a question the log cannot answer: not whether an exception was granted, but why it was justified, on what evidence, and under which version of the policy. At that point the answer has to be reconstructed from people and side channels, and its quality depends on who still remembers. What changes when agents make the decisions A human decision leaves traces in the world even when it is not logged: an email, an approval, a conversation. An agent decision does not. If the reasoning is not captured by the system taking the action, it does not exist anywhere. This is why the shift to agentic execution makes the distinction urgent rather than academic. When an audit log is the right tool Audit logs remain the correct instrument for security forensics, sequence reconstruction, and integrity monitoring of the systems themselves. Decision traces do not replace them; they sit above them and answer a different question. A well-run environment keeps both. FAQ Questions. Do decision traces replace our audit logging? No. Logs remain the right tool for sequence and integrity. Traces add the reasoning layer that logs were never designed to hold, and the two are complementary. Can we derive traces from existing logs? Only partially, and only for decisions whose rationale happened to be recorded. The evidence, policy version, and authority behind most past decisions were never captured, which is why traces are generated going forward rather than backfilled. Is this just a richer log format? The format matters less than when and by whom the record is made. A trace is produced by the deciding system at the moment of the decision, and signed so a third party can rely on it. That is a different guarantee from a log entry written by the system being audited. Start free Power up your business with the OriginTrail DKG. Create Knowledge Assets, build shared context graphs, enshrine decision traces. Open-source infrastructure. Community support. Zero cost. Go to OriginTrail → Go Pro The full Network Operating System. Custom agent frameworks (Hermes, OpenClaw, LangChain, Claude). Custom data pipelines. Custom integrations with your enterprise systems. Dedicated infrastructure and support. See nOS → --- # nOS vs Traditional Systems of Record URL: https://www.trace-labs.ai/compare/nos-vs-systems-of-record Glossary · Comparison nOS vs. traditional systems of record A system of record stores the state of business objects: accounts, orders, tickets, assets. nOS records the decisions made about those objects, with the reasoning that justified them, and enshrines the result so other parties can verify it. The two are complementary layers, not competing databases. Traditional system of record nOS Unit of record Objects and their current state Decisions and the reasoning behind them Answers What is true now Why this was allowed, and on what basis Boundary One organization, one vendor Multi-party by design Ownership of the record The vendor platform The organization, via an open knowledge graph Value to AI agents Data to read Precedent to inherit On vendor exit Migration project, history usually degrades Records remain, held independently What systems of record were built for Salesforce, SAP, ServiceNow and their peers solved a genuine problem: giving an organization one authoritative version of its objects. They were designed when the expensive thing was keeping state consistent, and they do that well. The reasoning that turned data into action was never their job, so it was left to people and, increasingly, is left to models. Where the gap shows up Ask any organization why a particular exception was approved last quarter and the answer lives in someone’s memory, a Slack thread, or a call that was never recorded. Deploy agents on top of that and they inherit the same blind spot: they can read the state but cannot see the judgement that produced it, so they cannot reuse it. Why this is not a replacement nOS does not store your accounts or your inventory, and it does not ask you to migrate. It connects to the systems already running and records the decision layer they omit. Framing it as a replacement misses the point: the object record and the decision record answer different questions and both are needed. FAQ Questions. Do we have to replace our CRM or ERP? No. nOS connects to the systems you already run. It records the decisions made around them rather than duplicating the objects they hold. Where does the data actually live? Private context stays in your infrastructure. What is enshrined in the OriginTrail DKG is what you choose to publish, typically the commitments and metadata that make a decision verifiable. What happens if we stop using nOS? The decision records remain, because they are held in an open decentralized knowledge graph rather than in a Trace Labs database. You can self-host and continue without data loss. Start free Power up your business with the OriginTrail DKG. Create Knowledge Assets, build shared context graphs, enshrine decision traces. Open-source infrastructure. Community support. Zero cost. Go to OriginTrail → Go Pro The full Network Operating System. Custom agent frameworks (Hermes, OpenClaw, LangChain, Claude). Custom data pipelines. Custom integrations with your enterprise systems. Dedicated infrastructure and support. See nOS → --- # Verifiable AI for Supply Chains URL: https://www.trace-labs.ai/solutions/supply-chain Solutions · Supply Chains Replaces · Siloed compliance databases · Vendor-controlled audit logs Verifiable AI for supply chains. Traceable handoffs from source to shelf. Global supply chains span dozens of suppliers, auditors, and regulators, yet the reasoning behind each compliance decision is scattered across siloed databases and vendor audit logs. nOS captures every audit rationale, exception approval, and handoff as a signed decision trace, so the "why" behind a shipment travels with it from source to shelf. Trace Labs works with networks like SCAN (the Supplier Compliance Audit Network) to make multi-party compliance verifiable and portable. Power up on OriginTrail (free) Need custom agents for supply chains? Go Pro. → Decision traces enshrined What gets captured Audit decision rationale Exception approvals across borders Customs and origin verifications Multi-tier supplier attestations Most relevant memory layers Shared & Verifiable Working memory captures private agent reasoning. Shared memory is where multi-party context coordinates. Verifiable memory is where decisions are enshrined. ~40% US imports audited via the protocol Deployment & partners SCAN, Supplier Compliance Audit Network Real enterprise data, processed in production, with decision traces enshrined as Knowledge Assets in the OriginTrail DKG. The problem The case for verifiable decisions in supply chains. Supply chain compliance is a multi-party problem solved with single-party tools. A retailer audits a tier-one supplier, that supplier attests for tier two, and a customs broker files on behalf of all of them. Each party keeps its own records, so when a regulator asks why a shipment was cleared, the answer has to be reassembled from email threads, PDF certificates, and a database nobody outside the company can inspect. Regulations have made that reassembly expensive. UFLPA forced-labor rules, the EU Deforestation Regulation, CBAM carbon reporting, and FSMA 204 food traceability all require evidence that reaches past your direct suppliers. AI agents deployed on top of the existing silos inherit the same blind spots: they can tell you a shipment was approved, not what evidence made approval defensible. How it works How a decision becomes verifiable. Step 1 · Capture An agent handling a shipment records the inputs it consulted: supplier attestations, audit reports, origin certificates, and the policy version in force at that moment. Step 2 · Share Context that other parties are entitled to see moves into shared memory, so the broker, the auditor, and the buyer work from the same facts instead of three reconciliations. Step 3 · Decide The agent applies policy and records the reasoning, including which exception was granted, who was entitled to grant it, and what precedent from earlier shipments it inherited. Step 4 · Enshrine The resulting decision trace is written to the OriginTrail DKG as a Knowledge Asset, so any counterparty or regulator can verify it later without access to your systems. What changes What this changes in practice. The practical change is where evidence lives. Instead of a compliance team assembling a defense after a question is asked, the reasoning behind each clearance already exists in a form a customs authority or buyer can check. Determinations stop being assertions backed by filing cabinets and become records with lineage. That also changes what an agent is worth. An agent that clears shipments without leaving a trace saves time and adds risk. An agent that clears shipments and enshrines why each clearance was permitted compounds: the hundredth decision inherits precedent from the previous ninety-nine, and the audit that follows is a query rather than a project. FAQ Questions about supply chains. How is this different from supply chain traceability software? Traceability software records where goods moved. A decision trace records why a compliance decision about those goods was allowed: the evidence consulted, the policy applied, the exception granted, and the precedent inherited. Movement data alone does not answer a regulator asking how a determination was reached. Do our suppliers have to adopt nOS as well? No. Suppliers can keep their existing systems. What matters is that attestations and audit evidence entering a decision are referenced and signed, so the resulting trace is verifiable even when the parties run different software. Does this expose commercially sensitive supplier data? No. Working memory and private context stay in your infrastructure. Only the commitments and metadata you choose to publish are recorded, which lets a counterparty verify that a decision followed policy without seeing your pricing or supplier list. Which supply chain regulations does this help with? Any regime that requires defensible evidence rather than a stored document: UFLPA and comparable forced-labor rules, the EU Deforestation Regulation, CBAM, and FSMA 204 traceability. nOS does not file for you; it makes the reasoning behind each determination auditable. How long does it take to see value? The first useful traces appear as soon as one decision type is instrumented, typically a single high-volume determination such as origin verification. Value comes from depth on one workflow rather than shallow coverage of many. What happens to our existing audit records? They remain valid and can be referenced as inputs. nOS does not require migrating history; it changes how decisions are recorded from the point of adoption forward. Start free Power up your business with the OriginTrail DKG. Create Knowledge Assets, build shared context graphs, enshrine decision traces. Open-source infrastructure. Community support. Zero cost. Go to OriginTrail → Go Pro The full Network Operating System. Custom agent frameworks (Hermes, OpenClaw, LangChain, Claude). Custom data pipelines. Custom integrations with your enterprise systems. Dedicated infrastructure and support. See nOS → --- # Verifiable AI for Cybersecurity URL: https://www.trace-labs.ai/solutions/cybersecurity Solutions · Cybersecurity Replaces · Single-vendor SIEM · Siloed threat databases Verifiable AI for cybersecurity. Threat intelligence that compounds across agents. Security teams run a growing fleet of detection and response agents, yet threat reasoning stays trapped in single-vendor SIEMs and siloed threat databases. nOS records detection rationale, escalation paths, and cross-team response decisions as shared, replayable traces, so intelligence compounds across agents instead of being re-derived. Every decision is independently verifiable, giving auditors and downstream teams a trustworthy record without vendor lock-in. Power up on OriginTrail (free) Need custom agents for cybersecurity? Go Pro. → Decision traces enshrined What gets captured Detection rationale and escalation paths Cross-team remediation decisions Threat-actor intelligence with full lineage Vendor-independent incident timelines Most relevant memory layers Shared & Verifiable Working memory captures private agent reasoning. Shared memory is where multi-party context coordinates. Verifiable memory is where decisions are enshrined. Multi-vendor No platform lock-in Portable Threat context travels with you Deployment & partners Security teams across regulated industries Real enterprise data, processed in production, with decision traces enshrined as Knowledge Assets in the OriginTrail DKG. Powering Umanitek’s Guardian internet safety agents. The problem The case for verifiable decisions in cybersecurity. Security operations already run on machine speed. Agents triage alerts, enrich indicators, and increasingly take containment actions. What they do not produce is a defensible record of judgement: which signals were weighed, which were dismissed, and on whose authority a host was isolated at 03:00. That gap becomes expensive under NIS2 and DORA, where incident reporting runs to fixed deadlines and supervisors ask how a classification was reached. It is also why threat intelligence rarely compounds: each vendor platform holds its own context, so the fiftieth investigation starts almost where the first one did. How it works How a decision becomes verifiable. Step 1 · Capture The triage agent records the detections, enrichment sources, and asset context it consulted, together with the detection logic version in force. Step 2 · Share Threat context that peers or a sector CERT are entitled to see moves into shared memory, so intelligence accumulates across teams instead of per vendor. Step 3 · Decide Containment and escalation decisions record the reasoning: severity rationale, the exception that permitted an out-of-hours action, and the precedent from similar incidents. Step 4 · Enshrine The trace is enshrined in the DKG, giving auditors and regulators an independently verifiable account of how the incident was handled. What changes What this changes in practice. The operational change is that judgement becomes an asset rather than a byproduct. Today an analyst decision lives in a ticket comment and evaporates when the ticket closes. Captured as a trace, it becomes precedent the next investigation inherits, which is what makes a security program improve rather than merely repeat. For leadership the change is evidentiary. When a supervisor or board asks why an incident was classified as it was, the answer is retrieved rather than reconstructed, and it holds up because it was recorded at the moment of the decision by the system that made it. FAQ Questions about cybersecurity. Does nOS replace our SIEM or EDR? No. nOS sits above detection tooling. Your SIEM and EDR keep producing signal; nOS records the decisions agents and analysts make on that signal, so the reasoning survives even if you change vendors. How does this help with NIS2 or DORA reporting? Both regimes ask how an incident was classified and what was done in response, on a deadline. Because the rationale is captured at execution time rather than reconstructed afterwards, the reporting narrative is assembled from traces rather than from memory and chat logs. Can we share threat intelligence without exposing our environment? Yes. You control what is published. Indicators and judgements can be shared while asset names, internal topology, and private context remain in your own working memory. What stops an agent from acting outside policy? Policy is applied at decision time and recorded with the action, so an out-of-policy action is visible as such rather than indistinguishable from a routine one. The trace shows which rule permitted the step and which precedent it relied on. Does this slow down response? No. The rationale is captured as part of the action rather than as a separate documentation step, which is precisely why it survives; documentation written afterwards is the part that gets skipped under pressure. How does this help a new analyst? They inherit precedent. Previous decisions on similar alerts are queryable with their reasoning intact, so the judgement of experienced responders becomes available rather than tacit. Start free Power up your business with the OriginTrail DKG. Create Knowledge Assets, build shared context graphs, enshrine decision traces. Open-source infrastructure. Community support. Zero cost. Go to OriginTrail → Go Pro The full Network Operating System. Custom agent frameworks (Hermes, OpenClaw, LangChain, Claude). Custom data pipelines. Custom integrations with your enterprise systems. Dedicated infrastructure and support. See nOS → --- # Verifiable AI for Rail and Transportation URL: https://www.trace-labs.ai/solutions/transportation Solutions · Transportation Replaces · Fragmented maintenance systems · Single-operator data silos Verifiable AI for transportation. Predictive maintenance powered by shared context across operators. Rail and transit networks depend on predictive maintenance, but maintenance context is fragmented across operators and single-vendor systems. nOS connects component provenance, maintenance events, and cross-border tracking into shared context graphs, so agents generate trusted insights from multi-operator data. Trace Labs works with operators such as Swiss Federal Railways (SBB) to make maintenance decisions verifiable across organizational boundaries. Power up on OriginTrail (free) Need custom agents for transportation? Go Pro. → Decision traces enshrined What gets captured Real-time component provenance Maintenance decision rationale Sensor anomaly investigation chains Cross-border tracking and recall lineage Most relevant memory layers Shared & Verifiable Working memory captures private agent reasoning. Shared memory is where multi-party context coordinates. Verifiable memory is where decisions are enshrined. Deployment & partners Swiss Federal Railways (SBB) Real enterprise data, processed in production, with decision traces enshrined as Knowledge Assets in the OriginTrail DKG. The problem The case for verifiable decisions in transportation. Rolling stock, track, and signalling assets outlive the organizations that maintain them. A component may be manufactured by one company, installed by another, serviced by a third, and inspected by a regulator that trusts none of their databases by default. Maintenance history is the asset, and it is scattered. Predictive maintenance agents make this sharper. A model can recommend deferring an intervention, but the operator carries the safety case. Without a record of the evidence and the authority behind that recommendation, the organization is accountable for a decision it cannot fully reconstruct. How it works How a decision becomes verifiable. Step 1 · Capture The maintenance agent records sensor history, inspection records, and the component provenance it relied on, along with the standard applied. Step 2 · Share Asset context that other operators and the entity in charge of maintenance are entitled to see moves into shared memory, so history follows the component rather than the company. Step 3 · Decide Deferral, inspection, and replacement decisions record their rationale, including the risk threshold used and the engineer who authorized any deviation. Step 4 · Enshrine The trace is enshrined in the DKG so a safety regulator, a leasing company, or the next operator can verify the maintenance reasoning independently. What changes What this changes in practice. The change is that maintenance reasoning outlives the contract and the database. When an asset is sold, leased, or transferred between operators, the receiving organization inherits not just a service history but the justification behind each intervention, which is what a safety case actually rests on. Internally it shortens the distance between a model recommendation and an accountable decision. The engineer who accepts or overrides a predictive recommendation is recorded alongside the evidence, so the organization can show a human exercised judgement rather than deferring to a model it cannot inspect. FAQ Questions about transportation. Who is this for within a rail or transit operator? Asset management, maintenance engineering, and safety teams that must justify interventions, plus the entity in charge of maintenance that carries regulatory responsibility for the asset. What happens when an asset changes operator? The decision history is attached to the asset rather than to one operator database, so the receiving organization inherits verifiable maintenance reasoning instead of a document dump. Does this work with our existing asset management system? Yes. nOS connects to the systems already in use. The asset register stays where it is; what nOS adds is the verifiable record of why each maintenance decision was made. Is Trace Labs working with transport operators today? Swiss Federal Railways (SBB) is named among the deployments where enterprise data is processed in production with decision traces enshrined as Knowledge Assets in the OriginTrail DKG. How does this interact with predictive maintenance models? The model produces a recommendation; nOS records what the recommendation rested on and who acted on it. That separation matters because models are retrained and retired while the safety case must remain explainable. Is this only relevant to rail? No. The same pattern applies wherever long-lived assets move between operators and regulators expect maintenance decisions to be justified, including transit fleets, aviation ground assets, and heavy logistics. Start free Power up your business with the OriginTrail DKG. Create Knowledge Assets, build shared context graphs, enshrine decision traces. Open-source infrastructure. Community support. Zero cost. Go to OriginTrail → Go Pro The full Network Operating System. Custom agent frameworks (Hermes, OpenClaw, LangChain, Claude). Custom data pipelines. Custom integrations with your enterprise systems. Dedicated infrastructure and support. See nOS → --- # Verifiable AI for Life Sciences and Healthcare URL: https://www.trace-labs.ai/solutions/life-sciences Solutions · Life Sciences & Healthcare Replaces · Opaque pharma data lakes · Unauditable trial pipelines Verifiable AI for life sciences. Patient data sovereign. Clinical evidence verifiable. Clinical and pharmaceutical decisions demand provenance, yet the underlying data sits in opaque lakes and unauditable trial pipelines. nOS links drug-interaction data, trial evidence, and regulatory submissions with full provenance, keeping patient data sovereign while making clinical evidence independently verifiable. Trace Labs collaborates with partners including BSI and Oxford PharmaGenesis on AI-ready, verifiable medical knowledge. Power up on OriginTrail (free) Need custom agents for life sciences & healthcare? Go Pro. → Decision traces enshrined What gets captured Drug interaction findings, full lineage Clinical trial context and amendments Regulatory submissions with reasoning Adverse event triage decisions Most relevant memory layers Working, Shared & Verifiable Working memory captures private agent reasoning. Shared memory is where multi-party context coordinates. Verifiable memory is where decisions are enshrined. HIPAA Aligned data flows EU AI Act Audit-ready by design Deployment & partners BSI · Oxford PharmaGenesis · Healthcare consortia Real enterprise data, processed in production, with decision traces enshrined as Knowledge Assets in the OriginTrail DKG. The problem The case for verifiable decisions in life sciences & healthcare. Clinical evidence is only as good as its provenance. A submission rests on data that moved through collection, cleaning, analysis, and interpretation, often across a sponsor, a CRO, and several sites. Regulators expect attributable, contemporaneous records, and reviewers ask what a given conclusion was based on. Introducing AI into that chain raises the bar rather than lowering it. If an agent summarizes safety signals or drafts a submission section, the organization must be able to show which sources informed the output and under what governance, especially as the EU AI Act brings obligations for high risk uses in health. How it works How a decision becomes verifiable. Step 1 · Capture The agent records the datasets, protocol version, and consent basis behind a piece of analysis, keeping patient level data inside the controlled environment. Step 2 · Share Findings and evidence that collaborators are entitled to see move into shared memory, so sponsor, CRO, and site work from one lineage rather than reconciled extracts. Step 3 · Decide Interpretation and inclusion decisions record their reasoning, including the criteria applied and any exception approved by a named reviewer. Step 4 · Enshrine The trace is enshrined in the DKG so evidence lineage can be verified by a regulator or partner without moving the underlying patient data. What changes What this changes in practice. The change is that evidence lineage becomes continuous rather than reconstructed at submission time. Instead of a team assembling provenance for a dossier under deadline, each analytical decision already carries its basis, which shortens the gap between science and a defensible regulatory narrative. It also makes AI usable in places where it currently stalls. Teams hesitate to let agents touch regulated evidence because they cannot show how an output was reached. When the agent records its sources and governance as it works, the barrier moves from principle to configuration. FAQ Questions about life sciences & healthcare. Does patient data leave our environment? No. Working memory and private context never leave your infrastructure. Only the commitments and metadata you choose to publish are recorded, which is what makes lineage verifiable without transferring personal data. How does this relate to HIPAA and GDPR? nOS is designed so that personal data stays under your control and disclosure remains your decision. That is the condition both regimes care about; the verifiable layer holds evidence about decisions, not the clinical records themselves. What does the EU AI Act change for us? It brings documentation, traceability, and human oversight obligations for higher risk uses. Decision traces address the evidentiary part directly, because the rationale is captured when the decision is made rather than reconstructed for an audit. Who is Trace Labs working with in this sector? BSI and Oxford PharmaGenesis are named among the collaborations, alongside healthcare deployments where decision traces are enshrined as Knowledge Assets in the OriginTrail DKG. Can this support a regulatory submission directly? It supports the evidence behind one. nOS does not produce regulatory documents; it makes the lineage of the data and decisions inside them verifiable, which is the part reviewers probe. How does this apply to real world evidence? Real world evidence draws on heterogeneous sources with uneven provenance. Recording which sources informed a conclusion, and under what governance, is what allows the conclusion to be defended later. Start free Power up your business with the OriginTrail DKG. Create Knowledge Assets, build shared context graphs, enshrine decision traces. Open-source infrastructure. Community support. Zero cost. Go to OriginTrail → Go Pro The full Network Operating System. Custom agent frameworks (Hermes, OpenClaw, LangChain, Claude). Custom data pipelines. Custom integrations with your enterprise systems. Dedicated infrastructure and support. See nOS → --- # Verifiable AI for Manufacturing URL: https://www.trace-labs.ai/solutions/manufacturing Solutions · Manufacturing Replaces · Fragmented MES · Disconnected quality systems Verifiable AI for manufacturing. Production intelligence that compounds on the factory floor. On the factory floor, quality and process decisions are spread across fragmented MES and disconnected quality systems. nOS captures quality decisions, process adjustments, and compliance checks as compounding knowledge, so production intelligence grows richer with every batch. The result is a verifiable record manufacturers can reuse across lines, suppliers, and audits. Power up on OriginTrail (free) Need custom agents for manufacturing? Go Pro. → Decision traces enshrined What gets captured Quality decisions across batches Process adjustment rationale Multi-tier supplier attestations Compliance events with full chain Most relevant memory layers Shared & Verifiable Working memory captures private agent reasoning. Shared memory is where multi-party context coordinates. Verifiable memory is where decisions are enshrined. Multi-tier Supplier traceability Real-time Quality decision capture Deployment & partners Manufacturing consortia · Industrial operators Real enterprise data, processed in production, with decision traces enshrined as Knowledge Assets in the OriginTrail DKG. The problem The case for verifiable decisions in manufacturing. A quality decision on the line is made in seconds and defended for years. When a nonconformance is dispositioned, the reasoning lives in an inspector’s judgement, a supplier certificate, and a specification revision that may since have changed. If a recall follows, scoping it means proving which units were affected and why each disposition was acceptable. Multi tier supply makes that harder, and the EU Digital Product Passport will make the evidence expectation explicit. Agents that inspect, disposition, and schedule can move faster than the record keeping around them, which is precisely the gap that turns a contained quality event into an uncontained one. How it works How a decision becomes verifiable. Step 1 · Capture The agent records inspection data, the specification revision in force, and the supplier attestations behind the material involved. Step 2 · Share Component and quality context that suppliers or customers are entitled to see moves into shared memory, so traceability survives the tier boundary. Step 3 · Decide Disposition decisions record the criteria applied, the concession granted, and who was authorized to grant it. Step 4 · Enshrine The trace is enshrined in the DKG, so recall scoping and customer claims rest on verifiable reasoning rather than reconstructed paperwork. What changes What this changes in practice. The change shows up first in containment. When quality decisions carry verifiable reasoning, the population affected by a defect is derived from the record rather than estimated conservatively, and the difference between a precise recall and a precautionary one is measured in production runs. Over time it changes supplier relationships. Attestations that are signed and referenced rather than emailed create a shared factual basis, so disputes about what a supplier certified become checks rather than negotiations. FAQ Questions about manufacturing. How does this change a recall? Scoping depends on knowing which units were affected and why each disposition was acceptable. When those decisions carry verifiable traces, the affected population is derived from records rather than estimated, which is the difference between a narrow recall and a broad one. Does this require replacing our MES or QMS? No. Those systems continue to run. nOS connects to them and adds the layer they were never designed to hold: the reasoning behind each decision, in a form other parties can verify. How does this relate to the Digital Product Passport? The passport regime expects product level evidence that follows the item across companies. Decision traces provide the provenance behind the claims a passport carries, rather than a static file that asserts them. Can suppliers contribute without our systems? Yes. Suppliers reference and sign their attestations; nOS does not require them to adopt your stack for the resulting trace to be verifiable. Where should a manufacturer start? With one decision type that is high volume and frequently questioned, usually nonconformance disposition. Instrumenting a single workflow deeply produces usable precedent faster than broad shallow coverage. Does this work across plants with different systems? Yes. Plants keep their own MES and QMS. Because traces reference signed evidence rather than requiring one shared database, the record remains verifiable across heterogeneous sites. Start free Power up your business with the OriginTrail DKG. Create Knowledge Assets, build shared context graphs, enshrine decision traces. Open-source infrastructure. Community support. Zero cost. Go to OriginTrail → Go Pro The full Network Operating System. Custom agent frameworks (Hermes, OpenClaw, LangChain, Claude). Custom data pipelines. Custom integrations with your enterprise systems. Dedicated infrastructure and support. See nOS → --- # Verifiable AI for Sport and Athlete Data URL: https://www.trace-labs.ai/solutions/sport Solutions · Sport Replaces · Vendor-locked analytics · Siloed scouting databases Verifiable AI for sport. Performance data with unbreakable provenance. Athletic performance and scouting data is typically locked inside vendor analytics and siloed databases. nOS gives federations and clubs performance analytics, scouting reasoning, and contract intelligence with durable provenance, owned by the organization and portable across seasons. Trace Labs works with bodies such as European Gymnastics and athletic federations to make sport data trustworthy and reusable. Power up on OriginTrail (free) Need custom agents for sport? Go Pro. → Decision traces enshrined What gets captured Athlete performance analytics Scouting and recruitment reasoning Contract and clearance decisions Anti-doping and integrity records Most relevant memory layers Shared & Verifiable Working memory captures private agent reasoning. Shared memory is where multi-party context coordinates. Verifiable memory is where decisions are enshrined. Athlete-owned Career data portable Federation-aligned Integrity-ready Deployment & partners European Gymnastics · Athletic federations Real enterprise data, processed in production, with decision traces enshrined as Knowledge Assets in the OriginTrail DKG. The problem The case for verifiable decisions in sport. An athlete career generates performance data, medical records, eligibility rulings, and anti doping history, held by clubs, federations, and service providers that change over time. The athlete rarely controls any of it, and when they move, the record does not reliably move with them. Governing bodies face the mirror image of the problem. Eligibility and sanction decisions must be defensible years later, often before an arbitration panel, and the evidence sits across organizations with different systems and different retention practices. How it works How a decision becomes verifiable. Step 1 · Capture The agent records the evidence behind an eligibility, selection, or medical clearance decision, along with the rule version in force at the time. Step 2 · Share Context the federation, club, or medical team is entitled to see moves into shared memory, under the athlete’s control where the data is personal. Step 3 · Decide The ruling records its reasoning: criteria applied, exception granted, and the official who authorized it. Step 4 · Enshrine The trace is enshrined in the DKG so the decision can be verified later by an arbitration panel without depending on one organization’s archive. What changes What this changes in practice. The change is that an athlete record becomes portable without becoming exposed. Clubs and federations verify that a clearance or eligibility ruling was properly made without receiving the underlying medical file, which is the balance that current data sharing arrangements rarely strike. For governing bodies it reduces the cost of being challenged. Because the rule version, evidence, and authorizing official are fixed at the moment of a ruling, defending that ruling years later does not depend on whoever still remembers the case. FAQ Questions about sport. Who owns the athlete data in this model? Ownership stays with the athlete and the institutions that create records. nOS records verifiable decisions about that data; it does not centralize the data itself, and disclosure remains controlled. Why does portability matter to a federation? Because eligibility and medical decisions depend on history. When the record follows the athlete verifiably, a receiving club or federation inherits context instead of restarting assessments. How does this help in a dispute? Arbitration turns on what was known and which rule applied at the time. A decision trace fixes both at the moment of the ruling, rather than relying on a later reconstruction. Is Trace Labs working with sports bodies? European Gymnastics is named among the federations and athletic organizations in this area, with decision traces enshrined as Knowledge Assets in the OriginTrail DKG. What about minors and sensitive medical data? Sensitive records stay in the controlled environment of the organization holding them. What is shared is verification that a decision followed policy, with disclosure controlled by the athlete or guardian where the data is personal. Can an athlete take their record when they transfer? That is the intent of the model. Because decision history attaches to the athlete rather than to a single club database, a transfer carries verifiable context instead of restarting assessments. Start free Power up your business with the OriginTrail DKG. Create Knowledge Assets, build shared context graphs, enshrine decision traces. Open-source infrastructure. Community support. Zero cost. Go to OriginTrail → Go Pro The full Network Operating System. Custom agent frameworks (Hermes, OpenClaw, LangChain, Claude). Custom data pipelines. Custom integrations with your enterprise systems. Dedicated infrastructure and support. See nOS → --- # Verifiable AI for Government URL: https://www.trace-labs.ai/solutions/government Solutions · Government & Critical Infrastructure Replaces · Centralized SCADA · Vendor-locked compliance systems Verifiable AI for government. Sovereign AI for energy, transport, and national security. Energy, transport, and national-security systems need decision-making that is both sovereign and accountable. nOS gives institutions verifiable decision traces with no vendor lock-in, full democratic accountability, and data portability by design. Trace Labs works with European agencies and national infrastructure operators to deliver sovereign AI built on the OriginTrail DKG. Power up on OriginTrail (free) Need custom agents for government & critical infrastructure? Go Pro. → Decision traces enshrined What gets captured Policy decisions with full lineage Cross-agency information sharing Critical infrastructure event chains Sovereign AI decision audit trails Most relevant memory layers Verifiable Working memory captures private agent reasoning. Shared memory is where multi-party context coordinates. Verifiable memory is where decisions are enshrined. Sovereign Owned by institutions, not vendors Audit-ready Democratic accountability Deployment & partners European agencies · National infrastructure operators Real enterprise data, processed in production, with decision traces enshrined as Knowledge Assets in the OriginTrail DKG. The problem The case for verifiable decisions in government & critical infrastructure. Public institutions are being asked to adopt AI while remaining accountable to citizens, auditors, and courts. A benefits determination, a procurement scoring, or a grid intervention must be explainable long after the model that informed it has been retrained or retired. Sovereignty compounds the requirement. Running decisions inside a vendor platform means the record of public reasoning lives somewhere the institution does not control, and the EU AI Act adds explicit documentation and oversight duties for high risk public uses. How it works How a decision becomes verifiable. Step 1 · Capture The agent records the evidence, the legal basis, and the policy version behind a determination, inside infrastructure the institution controls. Step 2 · Share Context other agencies are entitled to see moves into shared memory, so cross agency cases stop depending on document requests. Step 3 · Decide The determination records its reasoning, including discretion exercised and the official accountable for it. Step 4 · Enshrine The trace is enshrined in the DKG so oversight bodies and citizens can verify that process was followed, without the institution surrendering control of the underlying data. What changes What this changes in practice. The change is institutional memory. Determinations made under one administration, one policy version, and one supplier remain explainable under the next, because the reasoning was recorded independently of the system that produced it. It also alters the procurement conversation. When the evidentiary record lives in an open decentralized knowledge graph rather than inside a vendor platform, switching suppliers stops meaning losing the account of how public decisions were made. FAQ Questions about government & critical infrastructure. Can this run entirely on our own infrastructure? Yes. A self hosted deployment means no data leaves your environment, which is the usual condition for sovereign and critical infrastructure use. What does the EU AI Act require of public sector AI? Higher risk uses carry documentation, traceability, and human oversight obligations. Capturing rationale at decision time addresses the evidentiary core of those duties rather than producing documentation after the fact. How does this affect vendor lock-in? Decision traces are enshrined in an open decentralized knowledge graph rather than a vendor database, so the institution keeps the record even if it changes supplier. Does this make sensitive decisions public? No. What can be verified is that a decision followed policy and rested on stated evidence. The content of the underlying case data remains governed by the institution’s own disclosure rules. How does this support freedom of information requests? Requests turn on what was decided and on what basis. Records that already carry their evidence and policy version reduce the retrieval effort and the risk of an incomplete response. Is this compatible with sovereign cloud requirements? Yes. A self-hosted deployment keeps data inside the institution while still allowing decisions to be independently verified, which is the combination sovereign requirements usually demand. Start free Power up your business with the OriginTrail DKG. Create Knowledge Assets, build shared context graphs, enshrine decision traces. Open-source infrastructure. Community support. Zero cost. Go to OriginTrail → Go Pro The full Network Operating System. Custom agent frameworks (Hermes, OpenClaw, LangChain, Claude). Custom data pipelines. Custom integrations with your enterprise systems. Dedicated infrastructure and support. See nOS → --- # Verifiable AI for Construction URL: https://www.trace-labs.ai/solutions/construction Solutions · Construction Replaces · Paper-based building records · Disconnected compliance silos Verifiable AI for construction. Every compliance decision attached to the building’s digital twin. Building records are still largely paper-based and scattered across contractors, owners, and regulators. nOS attaches every compliance decision, material origin, and emissions record to the building digital twin, so context persists across the full lifecycle. Trace Labs contributes to initiatives such as the EU Digital Building Logbook to make construction compliance verifiable end to end. Power up on OriginTrail (free) Need custom agents for construction? Go Pro. → Decision traces enshrined What gets captured Material origin and emissions Compliance approvals, deviations Renovation history and energy data Sub-contractor attestations Most relevant memory layers Shared & Verifiable Working memory captures private agent reasoning. Shared memory is where multi-party context coordinates. Verifiable memory is where decisions are enshrined. EU CSRD Reporting-ready Deployment & partners EU Digital Building LogBook Real enterprise data, processed in production, with decision traces enshrined as Knowledge Assets in the OriginTrail DKG. The problem The case for verifiable decisions in construction. A building outlives every contract that produced it. Design decisions, material substitutions, and compliance sign offs are made by contractors and consultants who disperse at handover, leaving an owner with drawings that do not explain why the built asset differs from the design. Reporting has caught up with that gap. CSRD disclosure and the EU Digital Building LogBook expect evidence about materials and performance across the asset lifecycle, which is difficult when the reasoning behind each substitution was never recorded in a form anyone else can check. How it works How a decision becomes verifiable. Step 1 · Capture The agent records the specification, the material data, and the approval basis behind a design or substitution decision. Step 2 · Share Context the owner, contractor, and certifier are entitled to see moves into shared memory, so the as built record is assembled continuously rather than at handover. Step 3 · Decide Substitution and sign off decisions record their reasoning, including the standard applied and the party who accepted the deviation. Step 4 · Enshrine The trace is enshrined in the DKG and attaches to the building’s digital twin, so the reasoning survives the contract that produced it. What changes What this changes in practice. The change is that the as-built record assembles itself. Rather than a handover package compiled at the end by whoever remains, each substitution and sign-off is recorded with its justification as it happens, so the owner receives reasoning rather than only drawings. For the asset owner this compounds across the lifecycle. Refurbishment, resale, and disclosure all depend on knowing what was actually built and why, and that knowledge currently degrades every time a contract ends. FAQ Questions about construction. What problem does this solve at handover? Owners typically receive documents without the reasoning behind deviations. Verifiable traces mean the as built record explains why the asset differs from design, and who accepted each change. How does this support CSRD reporting? CSRD disclosures rest on material and performance claims. Traces provide the provenance behind those claims, so reporting is derived from verifiable records rather than assembled from contractor spreadsheets. What is the EU Digital Building LogBook connection? The Digital Building LogBook is named among the deployments in this area. The logbook is the container for lifecycle building data; decision traces supply verifiable provenance for what it contains. Does every subcontractor need to participate? No. Parties reference and sign the evidence they already produce. The value comes from decisions being traceable, not from every firm adopting the same software. Who benefits most, the contractor or the owner? The owner carries the asset for decades and gains most, but contractors benefit too: a signed record of why a deviation was accepted is the clearest defense against a later liability claim. Does this require BIM? No, though it complements it. BIM describes the asset; decision traces explain why the asset differs from the model and who accepted each change. Start free Power up your business with the OriginTrail DKG. Create Knowledge Assets, build shared context graphs, enshrine decision traces. Open-source infrastructure. Community support. Zero cost. Go to OriginTrail → Go Pro The full Network Operating System. Custom agent frameworks (Hermes, OpenClaw, LangChain, Claude). Custom data pipelines. Custom integrations with your enterprise systems. Dedicated infrastructure and support. See nOS → --- # Verifiable AI for Financial Services URL: https://www.trace-labs.ai/solutions/financial-services Solutions · Financial Services Replaces · Opaque audit logs · Vendor-controlled compliance pipelines Verifiable AI for financial services. Compliance with end-to-end decision traces. Financial compliance hinges on explaining why a determination was made, yet that reasoning hides inside opaque audit logs and vendor-controlled pipelines. nOS captures regulatory determination rationale and end-to-end decision traces that any auditor can independently verify. The result is compliance that is transparent, portable, and resistant to vendor lock-in. Power up on OriginTrail (free) Need custom agents for financial services? Go Pro. → Decision traces enshrined What gets captured Regulatory determination rationale Cross-border compliance decisions AML and KYC reasoning chains Independently replayable audit trails Most relevant memory layers Shared & Verifiable Working memory captures private agent reasoning. Shared memory is where multi-party context coordinates. Verifiable memory is where decisions are enshrined. Weeks → minutes Audit preparation Third-party Independently verifiable Deployment & partners Financial regulators · Compliance consortia Real enterprise data, processed in production, with decision traces enshrined as Knowledge Assets in the OriginTrail DKG. The problem The case for verifiable decisions in financial services. Financial institutions have long been required to explain decisions to supervisors. What has changed is who makes them. When agents assist in credit assessment, transaction monitoring, or client onboarding, the institution still owes a defensible account of why an outcome was reached and what governed the model behind it. The regimes are explicit. Model risk expectations, DORA operational resilience requirements, and the EU AI Act treatment of creditworthiness as a high risk use all assume the institution can produce evidence on demand. Audit preparation that takes weeks of reconstruction is a symptom of decisions that were never recorded with their reasoning. How it works How a decision becomes verifiable. Step 1 · Capture The agent records the data consulted, the model version, and the policy in force behind an assessment or alert disposition. Step 2 · Share Context that group functions or third parties are entitled to see moves into shared memory, so risk teams stop reconciling separate views. Step 3 · Decide The decision records its reasoning, including thresholds applied, overrides taken, and the person accountable for each override. Step 4 · Enshrine The trace is enshrined in the DKG so supervisors and internal audit can verify how an outcome was reached without a reconstruction exercise. What changes What this changes in practice. The change is that audit readiness stops being a project. When the reasoning behind assessments and overrides is captured at execution time, preparing for a supervisory review becomes retrieval rather than reconstruction, and the institution stops paying repeatedly for the same evidence. It also makes agent adoption defensible. Risk functions block agents they cannot explain; an agent that records its inputs, model version, and governing policy converts that objection into a control that can be reviewed. FAQ Questions about financial services. How does this shorten audit preparation? Because rationale is captured when the decision is made, evidence is queried rather than reassembled. The work shifts from reconstructing what happened to retrieving records that already exist. Does this apply to model risk management? Yes. Traces record which model version and which policy governed a given decision, which is the linkage model risk frameworks expect and the one most often missing in practice. What about third party and vendor risk under DORA? DORA focuses on resilience and oversight across providers. Traces enshrined in an open knowledge graph rather than a vendor system mean the institution retains the evidence independently of any single supplier. Is customer data exposed to other parties? No. Personal and commercially sensitive context stays in your infrastructure. Counterparties verify that a decision followed policy without gaining access to the underlying customer records. How does this differ from our existing audit trail? An audit trail records that a transaction or change occurred. A decision trace records why an outcome was permitted, including the policy, threshold, and override rationale, which is what supervisors actually ask about. Can this cover decisions made by humans, not agents? Yes. The same structure records human overrides and approvals, which matters because supervisory questions usually concentrate on exactly those interventions. Start free Power up your business with the OriginTrail DKG. Create Knowledge Assets, build shared context graphs, enshrine decision traces. Open-source infrastructure. Community support. Zero cost. Go to OriginTrail → Go Pro The full Network Operating System. Custom agent frameworks (Hermes, OpenClaw, LangChain, Claude). Custom data pipelines. Custom integrations with your enterprise systems. Dedicated infrastructure and support. See nOS →