Inveniam, the data infrastructure company that anchors more than $200 billion in private-market assets on-chain, has moved to close an industry gap with its latest partnership with Docugami.
Both parties have come together to bring verifiable, element-level document data to real-world assets.
Like most sectors, artificial intelligence has found its way into private markets; however, the data challenge also moved with it, as the data underpinning the assets being analyzed are sometimes trapped in unstructured documents that machines have never been able to read with precision.
The deal will see both Docugami’s Document Graph Markup Language (DGML) and Inveniam’s NVNM Chain in action.
The partnership arrangement followed earlier reporting where Inveniam announced that it plans to acquire $MANTRA, the regulated Layer 1 blockchain on which NVNM Chain was built as a Layer 2, in a transaction expected to close by the end of this month.
That deal followed a $20 million strategic investment Inveniam made in $MANTRA in August 2025, and the two firms jointly built NVNM Chain before Inveniam moved to bring the entire stack under one roof.
What is Docugami’s proprietary tech?
DGML was created by Jean Paoli, Docugami’s CEO, who is a co-author of the XML 1.0 standard and former president of Microsoft Open Technologies.
It converts leases, loan agreements, operating statements, and valuation reports into precisely labeled data elements.
Inveniam’s NVNM Chain then records those elements as tamper-evident, time-stamped artifacts on-chain, creating a cryptographic audit trail that any authorized counterparty can independently verify.
DGML is different from other document data technologies
Currently, there is already the tech to establish the source of a document as a whole, proving, for instance, that a specific rent roll existed at a given point in time.
DGML takes that innovation to the next level, as it makes it possible to verify individual data points extracted from within a document. For example, a rent figure from a specific lease clause, a loan-to-value ratio from an underwriting memo, or a net operating income line from an operating statement.
“The world’s most important business decisions are made on the basis of documents that machines have never been able to read properly,” Paoli stated, speaking about the Inveniam arrangement.
He said, “We have spent years building the technology to turn complex documents into data with unsurpassed precision. By opening DGML, we are inviting every participant in the private capital ecosystem to collaborate with us and build on a shared foundation.”
Docugami’s technology uses open-source large language models, small agentic reasoning models, and knowledge graph generation to convert complex business documents into structured, actionable data without requiring training data or templates.
Why do tokenized assets and AI agents need to anchor document data on-chain?
The tokenized RWA value has reportedly grown from around $14.1 billion in January to over $32.4 billion. However, the data quality problem still remains an issue, as tokenizing or attesting to a private asset is only as trustworthy as the underlying data.
Verifying the source of that data has required either trusting the issuer or conducting independent due diligence from scratch.
NVNM Chain, which Inveniam launched on May 7 as its purpose-built Layer 2 for private markets, was designed to serve as an attestation layer for agentic AI.
NVNM Chain records dataset exists and issues Proof of Origin, Proof of State, and Proof of Process attestations that trace what data drove any decision or transaction.
With DGML in the mix, AI agents will be able to read the document data and also make it both verifiable and audit-ready.
According to Patrick O’Meara, Chairman and CEO of Inveniam, “DGML is a foundational advance in how we read and structure the documents that drive private capital.”
He stated, “Anchoring DGML-extracted data elements on-chain is the natural complement: it ensures the data elements, once surfaced, can be trusted by every stakeholder who needs to use it.”
The two companies have not released a date when their solution will be publicly available.
