Docugami and Inveniam have announced that the Document Graph Markup Language (DGML) is now fully open and documented, with the complete specification, developer documentation, sample documents, and a Python open-source reference implementation available at www.dgml.io and GitHub under the Apache 2.0 license. The initiative aims to unlock the financial potential of verifiable document data by enabling element-level verification, traceability to source pages, and trustworthy inputs for AI agents.
DGML is now fully open and documented under the Apache 2.0 license, with a Python reference implementation and specification available at dgml.io.
The initiative addresses the trust gap in private-market asset valuation, where trillions of dollars depend on documents like leases, loan agreements, and appraisals.
DGML represents documents as graphs of precisely labeled, queryable data elements, tagged by domain meaning (e.g., "base rent," "liability cap").
Inveniam anchors DGML elements on the NVNM Chain, creating tamper-evident attestations such as Proof of Origin, Proof of State, and Proof of Process.
The standard enables element-level verification, traceability to source pages, consistent portfolio-wide querying, and trustworthy AI agent inputs.
Docugami and Inveniam invite the community to help shape the evolution of the format as an industry standard.
Docugami and Inveniam today announced that the Document Graph Markup Language (DGML) is fully open and documented, opening the way to a more accurate, powerful, and trustworthy document standard to unlock the massive financial potential of verifiable document data. The complete current specification, developer documentation, sample documents, and a Python open source reference implementation are available at www.dgml.io and github.com/dgml-io under the Apache 2.0 license.
"A huge percentage of the world's business information is contained in documents that were written for humans, not for computers, which is a big problem as AI agents become more vital in enterprise systems," said Jean Paoli, CEO of Docugami and co-creator of the XML document standard and former president of Microsoft's open source subsidiary. "Opening DGML will enable the creation of a transformative document standard where every chunk of document information can be named, traced to the exact page, and provable, so any document data can be trusted by an investor, an auditor, or an AI agent. We invite the community to help shape the evolution of the format. The implementations that emerge from real use cases are what strengthen and solidify a standard."
"DGML plus NVNM Chain makes trust a property of the data itself, not a promise from a platform," said Patrick O'Meara, CEO of Inveniam, the digital data infrastructure leader for private markets. "Together, we are creating the missing infrastructure for pricing, trading, and financing private-market assets, which has the potential to unleash trillions of dollars in economic activity."
Trillions of dollars in private-market assets – commercial real estate, private credit, infrastructure, funds – are valued on the basis of leases, loan agreements, operating statements, and appraisals. Despite the critical importance of the data extracted from these documents, there is no way for an investor, auditor, or counterparty to verify the source of any information or whether it has changed since the source document was executed. As the industry and AI agents take on this work – screening deals, monitoring covenants, updating valuations – the problem compounds: an unverifiable input becomes an unverifiable valuation.
DGML, a domain-native document format, represents any document as a graph of precisely labeled, queryable data elements, extracted without templates or extensive training. DGML tags describe what each element is in the document's own domain: a liability cap, a base rent, a NAV component, and how each element relates to each other in the document. Inveniam then anchors those individual elements on the NVNM Chain, its purpose-built receipts layer for agentic AI, creating tamper-evident attestations such as Proof of Origin, Proof of State, and Proof of Process. Together, DGML and NVNM Chain fundamentally transform the trust dynamic in financial workflows:
Element-level verification. An asset manager, auditor, or investor can confirm the provenance of a single figure – the rent escalation behind a cash-flow model, the covenant behind a credit decision – without exposing the entire document.
Traceability to the source page. Every element carries pixel-precise coordinates, so a valuation input resolves to a highlighted region on the executed lease.
Consistency across a portfolio. Documents of the same kind share one semantic vocabulary: a query that identifies one lease's expiration date can identify the expiration date across a thousand leases – the foundation for continuous, audit-ready NAV verification.
Trustworthy inputs for AI agents. Agents consume consistently labeled data elements instead of re-parsing prose, and any value they use can be independently checked against its on-chain attestation.
An open standard, not a proprietary dependency. Asset managers, valuation firms, fund administrators, auditors, lenders, and their technology providers can all build against one openly documented format, providing interoperable tooling and no vendor lock-in.
Docugami and Inveniam are applying the standard to live private-market workflows, beginning with valuation and NAV verification for hard-to-value assets: Docugami converts source documents into DGML natively, and Inveniam anchors the extracted elements on NVNM Chain, giving every stakeholder, from asset owner to auditor to AI agent, a cryptographic chain of custody from the number back to the page it came from.
Anyone can read the versioned specification, study annotated samples, and build with or contribute to the reference implementation. DGML's four independent layers – Semantic, Spatial, Attestation, and Readable – let implementers adopt only what they need; the same self-contained file can serve an LLM call, a browser view, and a compliance audit trail.
Docugami and Inveniam invite partners, customers, developers and others to participate in open collaboration to help strengthen the future standard and its open source reference implementation.
About Docugami
Docugami is the pioneer of business document intelligence. Founded by Jean Paoli, co-creator of the XML standard and former president of Microsoft's open source subsidiary, Docugami transforms complex business documents into precisely labeled, actionable data — without templates or training data — so organizations and AI systems can act on document data with confidence. Docugami is headquartered in Kirkland, Washington.
About Inveniam
Inveniam is the digital data infrastructure solution for private markets, empowering asset owners, managers, and service providers to credential, understand, extract, and deploy their data for any application. Inveniam's NVNM Chain serves as the receipts layer for agentic AI — hashing and anchoring verifiable data, including attestations such as Proof of Origin, Proof of State, and Proof of Process.
