Entrilia

The Clearwater shape at the small end, and it survives the Oxane refinement. The vendor's own positioning is that it builds the most advanced general ledger for private capital, and the accounting engine, the ownership builder for multi entity structures, the investor portal, the data room, the reporting layer and the waterfall connector all keep working with the models removed.

The company's own phrasing settles it: the platform is described as enhanced with agentic capability, which is an addition to a ledger rather than a description of one. Built rather than rejected because the artificial intelligence claims sit on the product surface, including a shipped protocol server, and not only in category essays, which is the distinction that rejected two other candidates in this same pocket.

An outcome asserted with no control named, which is the definition of this grade. The platform is said to be enhanced with agentic capability that proactively assists teams across accounting, reporting, data and workflows, and nothing states what that capability may do without a person, what it may write to, where an approval is required or what it is prevented from touching.

The gap is sharper than it would be elsewhere because of what this vendor has shipped: an interoperability protocol server that exposes a fund's live general ledger to whichever outside assistant the customer chooses, with nothing published about which permissions travel across that boundary. Two other vendors built from this same pocket ship the same kind of interface and both state explicitly that permissions and the audit trail travel with it. This one does not.

Nothing published on accuracy, validation, benchmarking, monitoring or drift. The closest thing is a single line offering traceable answers from live platform data, and it is held at C deliberately because it is ambiguous in context: the sentence sits in marketing about connecting an outside assistant, where traceable may mean the data is live rather than that an answer carries provenance back to a record.

That reading matters, because a comparable vendor in this pocket earned a B on this axis for the same idea stated consistently across its material as a property of every response. One phrase is not the same claim.

No named client, no case study, no quantified outcome and no client count located. The one named relationship is a technology partnership with a waterfall calculation vendor, announced with a quote from this company's own co founder rather than from a customer, so it evidences an integration rather than an adoption. Standard C: no named customer and self reported framing only.

Nothing published on training, tenant isolation, retention or what happens to fund data at termination. The omission is most consequential at the protocol boundary, since data leaves for an assistant the vendor does not control and nothing states what is sent, what is retained on either side, or whether the customer's own model provider terms govern it. The vendor has moved the model outside its own perimeter and published nothing about the perimeter.

No privacy position, processing terms, sub processor list, retention schedule or framework compliance statement located. Per contact access control on the investor portal is published, letting a manager limit which documents and data each investor contact can reach, and that is an access control rather than a privacy posture.

No certification, attestation or trust portal located. Two claims exist and neither is a credential. The vendor says customer data sits behind a leader in cloud security, naming no provider, which is the same borrowed authority pattern found elsewhere in this session where a standard or a supplier is gestured at rather than named.

And it describes a world class cybersecurity model with behavioural analytics detecting attacks and previously unseen exploits, which is a product capability claim rather than evidence that anyone has audited it.

No licence, no supervised test and no programme enrolment, which is the expected position for fund accounting software sold to private capital managers rather than to supervised deposit taking institutions.

Nothing published. Consumer protection exposure is among the lowest in the index, since the work is fund accounting, allocation and investor reporting and no decision about a person's access to a financial service is being made. Recorded for completeness.

No published allocation of responsibility and no route for an affected party. As with the other accounting platforms in this pocket the affected party is identifiable and is not the customer: an investor whose capital account statement is produced from the same ledger the models help populate, and who has no relationship with this vendor.

No model, provider, version or hosting arrangement named for the agentic capability or for the models said to power its input forms, which are described only as the latest in interfaces, artificial intelligence and machine learning. One genuine and unusual structural fact is published and does not lift the grade: for the protocol path the vendor supplies no model at all, since the customer brings whichever assistant they already run. That discloses the shape of one path and leaves the vendor's own inference entirely unaccounted for.

The strongest row on this record and unusually good for a company at this stage. Three concrete things rather than an architecture claim: the platform is the general ledger itself for the funds that run on it, so it is the system of record rather than a layer beside one; a named standard connector runs the waterfall calculations of a named third party engine inside the application; and a shipped interoperability protocol server exposes live platform data to outside assistants.

Positioning as open, extendable by end users and by developers, is backed by those specifics rather than asserted alone. Held off A on scale and breadth: one named connector, no partner ecosystem, no certified platform connectors and no published installed base for any of it to be deep into.

Cloud native is stated and that is the whole of the disclosure. No region, tenancy, residency or hosting arrangement is published by the vendor. An aggregator profile lists a United States server location, which is not creditable under the source test and is recorded only so the next pass does not mistake it for a vendor statement.

No price, tier, unit or pricing basis published, and no pricing page located. Worth noting only because a direct competitor built from this same pocket does publish a starting annual figure, so the absence here is a choice rather than a category norm.

Genuinely narrow, which is what this grade is for and why only a small minority of the index carries it. The buyer is the private capital fund and its administrator: private equity, venture capital, private debt and alternative asset managers, all inside one operational lane. No client count, no assets figure, no geographic footprint and no named institution is published, and the company is at seed stage with a single institutional investor.

Compare the other four vendors built from this pocket, which publish client counts in the thousands and assets in the trillions. This is not a criticism of the product, it is the axis measuring what it measures.