Nammu21

Reading the documents is the whole job. No two credit agreements are drafted alike, and the system recognizes the context and structure of provisions and maps the relational connections between them. The founder describes this as requiring a sophisticated large language model to transform provisions into interoperable digital data. Language processing and the proprietary loan language index, the NEL Protocol, are the product rather than a layer on it. Without them, what remains is a document repository requiring the manual reading this exists to replace.

The technology is described as fully automated in recognising the context and structure of complex loan documents, and the output becomes a digital security master feeding downstream operations, which is a consequential artifact to generate without a described check.

Nothing published sets out whether extracted provisions are validated before entering a security master, what confidence indication accompanies a parsed clause, or how a loan operations team reviews the digitisation of a document their institution is contractually bound by. A planned agent layer would extend automation further with no stated controls.

The named protocol is the transparency mechanism, because a loan language index that classifies provisions into interoperable identifiers is inspectable by construction: a user can see how a clause was categorised rather than receiving an opaque score, and interoperability is the stated goal so the taxonomy must be legible to both machines and people. The public database also lets a prospective buyer examine output quality directly against filings they can pull themselves. What is missing is measurement: no extraction accuracy figures, no error analysis, no model documentation and no stated validation support.

The investor list is institutional, and strategic rather than financial. An exchange operator's venture arm, a global custodian bank and a Swiss global bank all invested in one round, and the existing register includes another global bank and two digital asset focused funds. Institutions putting capital into a loan data standard signal an intent to use it.

A named integration with the dominant syndication platform describes Nammu21 as producing authoritative golden source data for that market. The public database holds more than 5,000 loans from filings since 2018. What is absent is customer specificity. The company describes work with several leading institutions without naming one or quantifying an outcome.

Data provenance is clean and clearly separated. The public database is built by programmatically isolating and extracting loan agreements from public filings, so its entire basis is disclosed material a user could verify independently. The company describes the extraction as its own technology rather than licensed content. The private side is handled per institution. What is not addressed is the boundary between them.

Nothing states whether provisions parsed from a client's private credit agreements inform the language index that serves other clients, which matters when the product's value grows with every document it reads.

The privacy exposure is structurally light. The subjects are corporate borrowers, lenders and facilities rather than individuals, and a substantial part of the data comes from public regulatory filings that are already disclosed. The confidentiality question that remains is commercial, since private credit agreements are unfiled and contain negotiated terms that lenders treat as sensitive. No data handling framework, retention schedule or subprocessor list is published to govern that private side.

Nammu21 publishes no trust center, enumerated certification list, attestation scope or audit period. Strategic investment from a global custodian, a Swiss global bank and an exchange operator implies each conducted diligence. Institutions placing private credit agreements on the platform would also require assurance before doing so. The control environment is very likely stronger than its public disclosure shows.

Nammu21 supplies technology and holds no licence, which is expected, and its regulatory engagement is thinner than the domain would support. Syndicated lending carries securities law questions around information sharing between lenders, the public database is assembled from securities filings, and the stated ambition to create digital credit instruments and securities tradeable on chain raises questions about what such an instrument is under securities law. None of that is addressed publicly, and no supervisory instrument is named as a design target.

The subjects are documents and corporate facilities rather than people, so this reads as accuracy governance, and the stakes are legal rather than statistical: a misread covenant, incorrectly mapped commitment or wrongly extracted pricing provision propagates into a security master that operations, agents and lenders then rely on, and the underlying agreement remains the binding instrument regardless of what the digitisation says. Nothing public reports extraction accuracy, identifies which provision types or drafting conventions the system handles least reliably, or describes reconciliation against the source document.

One property helps, and it is structural rather than promised. The source documents are the binding legal instruments and much of the public database comes from filings, so any extracted provision can be checked against an authoritative original that the vendor did not produce. That makes errors findable, which extraction products often cannot offer. Nothing binds the vendor beyond that. There is no accuracy guarantee, no remediation term where a wrongly digitized covenant flows into a security master, and no published error rate.

Provenance is disclosed on both halves of the chain. The analytical layer is proprietary and named, built around the company's own loan language index, the NEL Protocol, and described as its own large language model work rather than a wrapper. The data layer for the public database is explicitly sourced from public regulatory filings. A buyer knows both who built the intelligence and where the data came from. What is not published is the infrastructure beneath, with no model or hosting providers named and no subprocessor list identifying who processes private credit agreements.

The consequential integration is with the dominant syndication platform in this market, which describes Nammu21 as digitizing credit agreements into authoritative golden source data for its own users. The output therefore reaches lenders and buy side firms through software they already operate, without separate adoption. Interoperable digital identifiers are the design premise, so data is built to travel between systems rather than sit in one. Breadth is the gap. No loan operations, agency or portfolio systems are named individually beyond that partner, and there is no public developer documentation.

Delivery is cloud hosted, with a public database available directly and an enterprise platform for institutions. The loan market it serves is global, even though the public database draws on domestic filings. Hosting regions, residency options and transfer mechanisms are not published, and there is no public subprocessor list. Nor is tenancy separation described between institutions whose private credit agreements sit on the same platform.

Enterprise rates are unpublished, but the public database has a stated free and premium structure with a trial available. A prospect can use part of the product and see what the data looks like before any sales conversation. The enterprise platform's pricing basis, whether by documents processed, facilities managed or seats, remains undisclosed.

Buyers span both sides of the syndicated market, covering agent banks and lenders on the sell side and private credit funds and buy side institutions on the other, with the platform also serving legal professionals and equity analysts through the public database. Instrument coverage runs across syndicated and bilateral loans and private credit facilities. The boundary is deliberate and narrow: this is the corporate loan market only, addressing a multi trillion dollar segment with nothing for consumer credit, structured products, payments or insurance.