How DMaaST is connecting value chains with the Decentralized Knowledge Graph, and what it could mean for the factories of the next decade
Manufacturers are investing heavily in digital twins, AI-driven planning, and sustainability reporting, and are preparing for the EU’s Digital Product Passport (DPP). All of these depend on data that crosses company boundaries: supplier materials, production events, logistics records, and environmental footprints, each held in different systems and formats. That data is usually fragmented, hard to combine, and impossible to verify once it leaves its source.
DMaaST tackles this at the foundation. The project’s Smart Manufacturing Platform is built on a shared data layer powered by the OriginTrail Decentralized Knowledge Graph (DKG). We are testing it with JPB Système in aeronautics and Kamstrup in electronics.
What is the Decentralized Knowledge Graph
A knowledge graph stores information as connected facts: this meter was assembled on this line, from this component batch, with this energy use. The DKG adds two additional things to this:
- It is decentralized. No single company or platform owns the network or holds everyone’s data. Each organization publishes and controls its own data.
- It is verifiable. Each record is published as a Knowledge Asset: a structured package of information with a unique identifier and a cryptographic fingerprint anchored on a blockchain. Anyone can check where a record came from and that it has not been altered.
The DKG is open source, built on established W3C and GS1 standards, and has already been used in sectors such as food, rail, and pharmaceuticals.
Five ways the DKG improves manufacturing
1. Partners work together without adopting one system.
Value chains are built from companies that will never run the same software. Instead of forcing a common schema, the DKG maps each partner’s data to shared vocabularies, such as GS1 EPCIS. In DMaaST, production data from France and Denmark becomes comparable without either partner replacing its existing systems.
What this means for manufacturers: fewer one-off integrations and faster onboarding when a new supplier or customer joins.
2. Trust without a middleman.
Today, supplier data is trusted mainly because of who sent it. With Knowledge Assets, trust attaches to the data itself. Buyers, auditors, or regulators can verify origin and integrity independently, while the data owner controls who sees the content.
What this means for manufacturers: lower audit effort, fewer disputes over whose numbers are right, and data claims you can defend.
3. Traceability reaches the individual product.
Knowledge Assets carry recognized identifiers: GTINs, GLNs, and serial-level identifiers for individual items, so a specific unit can be followed for quality investigations, recalls, warranties, and repair. This is also what a DPP needs. The EU’s DPP registry went live in July 2026, and sector requirements will follow.
What this means for manufacturers: passport readiness becomes a side effect of good data infrastructure.
4. Sustainability data reflects what actually happened.
Most product footprints are averages. On the JPB and Kamstrup production lines, the consortium is linking lifecycle inventory data (energy, materials, emissions) to the traceability events of individual serialized products. The aim is a footprint based on each unit’s real production history.
What this means for manufacturers: environmental claims that hold up under scrutiny.
5. AI and digital twins work from data you can check.
In DMaaST, the DKG is the data layer for the platform’s cognitive digital twins and AI-powered multi-objective decision support system. These help anticipate disruptions and balance production, logistics, and stock. Every recommendation can be traced back to its source data. Looking ahead, AI assistants and agents can use the same graph to retrieve verified facts and cite where they came from.
What this means for manufacturers: AI recommendations engineers can question, audit, and trust.
Open by design
The DKG’s open-source code and open standards also protect against vendor lock-in. Data isn’t tied to a single vendor, and no platform operator owns the network. Investment in data quality keeps its value whatever tools come next. This fits the direction of European initiatives on data sovereignty and industrial data sharing.
Three learnings to consider:
- Start with identifiers. If part numbers, serial numbers, and location codes don’t line up across systems, no graph or AI can fix it. In DMaaST, linking lifecycle data to production records starts with matching these identifiers across partners’ systems. That alignment is the first piece of work, not an afterthought.
- Verification changes behavior. When partners know shared data can be checked, discussions shift from “whose data is right” to “what does the data tell us.”
- AI amplifies data quality, good or bad. Digital twins and decision tools inherit every gap and error in their inputs. Data that can be verified is the cheapest form of AI risk management.
The foundation comes first
Smart manufacturing is often described in terms of what sits on top: twins, dashboards, AI copilots. DMaaST’s experience points the other way. The factories that benefit most from these tools will be the ones that first make their data connected, verifiable, and shared on terms they control. The DKG does exactly that, and the passport, the footprint, and the AI recommendation all follow from it.
Ana Bevc, Head of Operations @Trace Labs — OriginTrail Core Developers
