Alkymi

The removal test leaves the manual document processing that is the operational bottleneck for every private markets investor. The chief executive states the founding decision plainly, that what the company got right from the beginning was putting machine learning at the core of the system, and the current stack names three model classes working together: machine learning, large language models and agentic components, engineered specifically for investment workflows. A separate generative product enriches datasets and generates insight from the same documents. Nothing here is a rules layer with models attached.

One capability functions as genuine oversight rather than as extraction: for each fund the platform monitors whether all expected data has been received and is complete, so the system checks what is missing rather than only processing what arrived. That matters because in private markets the dangerous failure is silence, a report that never came, and no parser catches an absent document.

Validation is named alongside integration and analysis in the company's own description of the workflow. Output is structured data for investment teams to act on rather than a decision. What is absent is any description of confidence handling on individual extracted values or a review step before figures reach downstream systems.

Two controls are described rather than implied. Validation is named as a distinct step in the workflow alongside integration and analysis, and completeness monitoring per fund checks the integrity of the input set itself, which addresses the failure mode that silently corrupts downstream analysis. Improved data quality is stated as an outcome customers rely on.

What is missing is measurement of the extraction itself: no accuracy rate, error figure or reconciliation statistic is published, and the headline claim of unlocking all of the data trapped in unstructured documents describes coverage rather than correctness, which are different things when the output feeds valuations.

Five institutions are named, spanning a major United States insurer, a large Dutch pension manager, an American investment group, an Australian superannuation fund and a leading investment management software provider. The aggregate figure is the largest recorded anywhere in this index: investment managers representing more than 20 trillion dollars in assets, including top pension funds, global banks, major hedge funds and sovereign wealth funds.

Two of those named customers also invested in the August 2025 financing round alongside two venture firms, which is the customer investor pattern appearing twice in one transaction. The company was named to a well known financial technology hundred list in 2024 and has two named partnerships extending its reach.

No data boundary statement was located. Extraction models improve with exposure to more manager specific formats, and this platform processes documents on behalf of institutions holding more than 20 trillion dollars in assets, so the format learning available from that volume is substantial and directly transferable between clients.

The documents also arrive under confidentiality from general partners to their limited partners, and the platform sits inside that relationship without being party to it. Nothing states what is retained, whether learning is shared across clients, or how competing allocators' portfolios are separated.

Structurally favourable by subject matter, since the payload is institutional investment documentation rather than personal data, so consumer privacy questions do not arise. What the platform holds instead is the complete private markets position of pension funds, sovereign wealth funds and insurers, which is commercially sensitive at the highest level even though no individual appears in it. Held at B because no data processing terms, retention schedule or subprocessor list was located, and clients span United States, European and Australian regimes.

No attestation, certification, trust centre or enumerated framework was located. Institutions representing more than 20 trillion dollars have completed vendor assessment on this platform, including pension funds and sovereign wealth funds whose supplier requirements are among the most demanding anywhere, so the assurance exists privately and none of it is published for the next buyer.

No supervisor, statute, instrument or reporting standard is named. The gap widens with the private credit product, because credit risk monitoring output feeds valuation and provisioning judgements at regulated asset managers and lenders, and because private markets reporting operates against established industry disclosure templates that determine what investors receive. None of that framework is identified.

No individual is assessed and two adapted exposures apply. The first is shared with every extraction platform in this lane: parsing quality tracks document quality, so a large institutional manager producing standardised reporting is read more reliably than a smaller or first time fund, which means an investor's view of its least transparent holdings is its least reliable.

The second is new and comes from the credit product, since early detection of deteriorating facilities means a model flags a borrower as weakening before any covenant breach, and that borrower is a private company that never sees the assessment, cannot contest it, and may find its access to capital affected by a signal derived from documents it supplied for another purpose.

No guarantee, indemnity or falsifiable accuracy commitment was located. The completeness monitoring gives a client a means of detecting one class of failure, missing data, which is more than most extraction platforms offer, and nothing describes what happens when an extracted value is simply wrong: no correction process, no notification path to clients whose reporting rested on it, and no statement of what is owed. The exposure extends beyond the direct customer because the platform's output flows into partner products and into asset servicers' own client reporting.

The model layer is described by class rather than by provider, naming machine learning, large language models and agentic components as the technologies engineered for investment workflows, which tells a buyer the architecture without identifying who supplies any of it.

No document source or portal provider is named on the ingestion side, which is a gap relative to comparable platforms that identify the investor portals they connect to, and no subprocessor list or hosting arrangement was located.

Integration is treated as a product surface rather than a feature, with pre built and custom connections delivering real time structured portfolio data into a client's downstream systems, and automatic retrieval and ingestion handling the upstream side.

Two named relationships extend it materially: a strategic partnership with a cloud native investment data management provider produces an integrated credit risk monitoring solution combining that firm's data infrastructure with this platform's document ingestion, and a consultancy partnership delivers joint implementation. A major investment management software provider is simultaneously a customer and an investor, which is the deepest form of platform alignment available.

No hosting provider, region selection, residency commitment or private deployment option was located. Clients include European pension managers and an Australian superannuation fund alongside United States institutions, each operating under different expectations about where investment data may be processed, and nothing published addresses it.

No pricing, packaging or basis of charge was located. The product now spans a core platform, a generative enrichment tool and a separately launched private credit solution, which is a structure that would ordinarily price by module, and nothing indicates whether charge scales with document volume, funds monitored, users or assets. Reported total funding also differs materially between sources, from 26 million to more than 35 million dollars.

The buyer set spans institutional investors, wealth managers, asset owners, asset servicing firms, pension funds, global banks, hedge funds, sovereign wealth funds and lenders, which is close to the whole institutional market rather than one segment of it.

Document coverage is the second dimension and it is enumerated specifically, running from capital calls and schedules of investments through loan agent notices and offering memoranda to brokerage and financial statements, so the platform handles the full range a private markets operation receives. Asset class coverage extends across private equity, alternatives and, through a purpose built product, private credit.