Oxane Partners

Without the document to data pipeline, the system would have no data in it at all. The firm's own diagnosis of its market is that the information needed to monitor a private credit investment sits in credit agreements, borrower reports, spreadsheets and email rather than in standardized feeds. Extraction is therefore not a convenience layer over a working system. Extraction, classification, spreading and covenant capture carry that throughput at $1.5 trillion of assets.

The firm sells a platform and people model in which staffed specialists do a material share of the work. Portfolio management, facility administration, valuations and reporting are conventional software beneath the extraction layer.

The firm staffs the human layer itself. The chief executive is on record that adoption in this market is anchored by governance, explainability and human oversight, and that the technology enhances workflows rather than replacing judgment. The platform and people model places the firm's own private credit specialists alongside the software in the delivery path. That staffing is presented as a service delivery model rather than as an adjudication layer over model output. Nothing describes what a specialist checks, when an extraction is escalated, or what confidence threshold routes a document to review.

Explainability is named as a principle by the chief executive and nothing documents it. For a product whose central function is extracting terms, covenants and financials from legal documents, the decisive model metric is extraction accuracy and no figure is published, nor any validation methodology, error rate, confidence scoring, human review rate or benchmark.

A wrongly extracted covenant threshold or maturity date propagates silently into monitoring, valuation and reporting, which makes the absence of any accuracy disclosure the most material gap on this vendor.

The scale claims are large and not one institution is named. Aggregate client assets under management on the platform are reported at over $1.5 trillion, rising through published milestones from one trillion, against more than a hundred clients. The tier claims are specific: twenty three of the top thirty global investment banks, thirteen of the top thirty private debt firms, ten of the top thirty institutional asset managers. All of it is self reported and unverifiable.

No client is identified anywhere, no case study attaches an outcome to an institution, and the executives quoted are the firm's own founders. The firm publishes original survey research on the asset class, which is credible thought leadership but is not customer evidence.

Unanswered, and the question is acute for this firm. It ingests executed credit agreements and facility documentation from twenty three of the thirty largest global investment banks and thirteen of the thirty largest private debt firms. That is among the most commercially sensitive document sets in finance, and competitors sit side by side on one platform. Nothing states whether documents or extracted terms from one client inform models or benchmarks available to another, how long documents persist, or whether any third party model provider sees them.

No published position on retention, deletion, subprocessors or cross border transfer, despite a delivery model that necessarily involves staff in one jurisdiction handling documents belonging to institutions in others. The data concerned is commercially confidential rather than consumer personal data for the most part, which changes the character of the risk without removing the disclosure gap.

No certification, attestation, security page or trust portal is published. That absence is conspicuous for a firm holding executed credit agreements, covenant terms, borrower financials and valuation data covering $1.5 trillion of assets for twenty three of the thirty largest global investment banks. All of those banks run vendor security assessments as a condition of onboarding. A firm at this scale almost certainly holds attestations that are shared under diligence rather than published.

No license or supervised status is published. The firm provides independent valuations, loan and agency services and facility administration, functions that in some jurisdictions attract registration or professional standards obligations. It claims no authorization, and its clients remain the regulated parties.

Beyond the principle stated in market commentary, nothing is published. No governance framework, model inventory, testing regime or accountability structure is described. Fairness in the consumer sense is largely inapplicable to a platform whose subjects are institutional borrowers and facilities rather than individuals. Governance disclosure still applies, and none is offered.

No published position covers responsibility for an incorrect extraction, valuation or covenant classification. Every affected party is a sophisticated institution operating under a negotiated contract, so recourse collapses into commercial terms rather than a consumer complaint route. That still leaves the question of who bears a valuation error unaddressed in anything published.

No model, provider or version is named behind the extraction, classification or generative drafting capabilities. Generative AI is referenced for producing memos, commentary and reports with no statement of what produces them. The inputs are confidential executed credit agreements, so a client would reasonably want to know whether those documents reach an external model provider at all.

Integration here means absorbing the unstructured estate rather than connecting to systems of record, and that is done well: the platform ingests credit agreements, borrower reports, spreadsheets and email and normalises them into a single portfolio view, unifying fund finance with portfolio management so leverage effects on returns and liquidity can be seen together, and it is modular so a client can scale coverage over time.

No fund accounting, custody, order management or administrator platform is named as a supported integration, which for a system sitting between managers, lenders and limited partners is the connection a buyer would want listed.

Undocumented. No deployment model, hosting arrangement, cloud provider, region or residency commitment is published. For this firm the question is substantive rather than formal. It processes confidential credit agreements and borrower financial reporting for institutions across Europe, North America and Asia from delivery centers in India. Where documents are held and processed is a live diligence question for every client, and it is left unanswered.

Nothing published. No pricing, tiers or bands, and no indication of how the platform and people model is charged, which is the specific question a buyer would have here since the offering deliberately blends licensed software with staffed specialist teams and those are normally priced on entirely different bases.

More than a hundred clients and over $1.5 trillion of aggregate client assets under management run on the platform. Customer composition is stated at tier level: twenty three of the thirty largest global investment banks, thirteen of the thirty largest private debt firms and ten of the thirty largest institutional asset managers, plus pension funds and sovereign wealth funds. Asset class coverage is equally wide.

It runs across direct lending, asset based finance and securitized products, and across commercial real estate, fund finance and infrastructure debt. The firm operates from four offices across Europe, North America and India.