Finastra

The Clearwater precedent, and the clearest instance of it in the index because the boundary is drawn so deliberately. Strip every model and Loan IQ still services the majority of the world's syndicated loan volume, LaserPro still generates loan documents as it has since 1986, Mortgagebot still originates mortgages, Trade Innovation still processes trade finance and the payments hub still clears payments.

The learned line is real, named and separately marketed rather than decorative, which is what makes this a build: seven distinct artificial intelligence products with their own factsheets, a partnership supplying document extraction into the loan system of record, and a named chief artificial intelligence officer. What it is not is the product.

The vendor's own framing is that intelligence works alongside its lending systems to surface insights, and every capability named is an assistant, a trainer, a search interface or an ingestion tool positioned beside the systems of record rather than inside them.

Gates are asserted in the vendor's own words and the product scope backs them up. The conversational assistant is described on the product page as supporting operational tasks and improving workflows with humans in the loop, the framing throughout is that intelligence works alongside lending systems to help teams make informed decisions, and every named capability is bounded to assistance, search, training, translation, document classification or ingestion.

None of them executes a transaction or issues a decision. Held at B and not A on the standing bar: no threshold, confidence band, default configuration, escalation path or approval requirement is published anywhere, so the phrase in the loop names the posture and not the mechanism.

One specific gap is worth stating because it is where an error would actually propagate: partner supplied document automation extracts credit agreement terms into the loan servicing system of record for a large share of the world's syndicated lending, and nothing describes what verifies an extracted term before it becomes the basis of an interest calculation on a live facility.

Nothing, and the absence is more conspicuous here than on most of this roster because the company has gone further than any peer on artificial intelligence organisationally. It employs a named chief artificial intelligence officer who is quoted on its own product page, publishes a white paper on generative artificial intelligence in financial services, and ships seven separately branded capabilities.

Against that: no accuracy or error rate for document classification, term extraction or translation, no validation methodology, no backtesting, no drift monitoring, no versioning, no external assessment, and no description of how any output is checked. The extraction accuracy question raised across this roster applies directly and is unanswered.

Named institutions and no attributed number. ING is carried as a named case study on the lending integration layer, Maybank on the trade platform and United Bank in a recorded interview. Alongside them sit installed base figures that are genuinely striking rather than promotional: twenty one of the top twenty five syndicated lenders, nine of the top ten global agent banks, roughly seventy percent of global syndicated loan volume, and more than three thousand institutions on the documentation product.

A customer is quoted saying almost ninety five percent of the portfolio now sits on the platform, and the quote is unattributed. Held at B on the line applied five times on this roster: named customers plus scale figures is B, and A needs a result attached to a name. Worth noting for the wider finding on this axis, since a vendor whose customer list is this senior publishes case studies without numbers.

Silent. Nothing states whether customer data trains or tunes any model, whether anything is pooled across institutions, what the conversational assistants retain from a session, or what happens to credit agreement content passed through the extraction flow.

The question has unusual force for this vendor because the material at stake is loan documentation and portfolio position data belonging to competing lenders on a shared platform, and because a third party model provider now sits inside one of the flows, which raises the same second order question that almost nothing in this index answers: what the supplier is bound to, not only what the vendor is bound to.

A general website privacy policy, a cookie policy and a legal library, and nothing at product level. No retention schedule, no deletion terms, no subprocessor list, and no tenant separation statement for the hosted lending service.

The unaddressed surface is specific rather than generic here: credit agreements are among the most commercially sensitive documents a corporate borrower produces, and a partner supplied extraction service now reads them as part of the onboarding flow, with nothing published about what either party retains of that content.

Nothing current or company wide was found. No certification is named on the lending or artificial intelligence product pages, none appears in the footer, and no customer facing trust portal surfaced. The single certification located is an announcement from 2018 covering one Swiss service bureau, which is entity scoped rather than group wide and eight years old, and under the standing credential test a narrowly scoped historic certification of one operating unit does not read across as a current credential for the platforms graded here.

Recorded as a queued question rather than a silent refusal, because this is unproven absence rather than evidenced absence: a vendor of this size selling to tier one banks certainly answers security questionnaires and very likely maintains a customer facing trust portal reached through its support or legal area, and that is where this grade would move. The gap is worth naming plainly in the meantime, since the platform holds loan positions and borrower data for most of the global syndicated lending market.

A software vendor with no licence, registration or supervised standing of its own. It sits closer to market infrastructure than almost any firm in this index, given the share of global syndicated lending serviced on its platform, but that is a concentration of operational importance rather than a regulatory position.

The documentation product carries regulatory content, periodic updates, a fifty state attorney network and a warranty, which is a substantive compliance offering and still a service to the buyer rather than standing held by the firm.

No published governance position, no fairness testing, no protected characteristic treatment, no named framework and no statement on any artificial intelligence regulation. The sharp point is organisational rather than technical: this vendor pairs a named chief artificial intelligence officer with no published artificial intelligence governance framework at all.

That mirrors the pattern recorded at Abrigo, which sells governance advisory and publishes three paragraphs, and at Prometeia, which sells model validation and publishes no evaluation of its own model. Fairness exposure is lower than for peers because the capabilities are assistive and none scores a borrower, which is recorded as mitigation and not as disclosure. The mortgage origination and documentation products reach consumer lending directly and nothing addresses that path.

No recourse position published. The exposure differs in kind from most of this roster and is arguably larger. Because the capabilities are assistive rather than decisioning, the risk is not a wrongly declined borrower but a wrongly captured instrument: an extracted margin, maturity, covenant or repayment term that enters the servicing platform incorrectly and is then applied across a syndicate.

Nothing describes who is responsible when that happens, how it is detected, or how liability divides between the vendor, the partner supplying the extraction and the agent bank operating the facility. The documentation product carries a stated warranty on regulatory content, which is the only liability commitment found anywhere in the estate and does not extend to model output.

The closest any vendor on this roster has come without arriving. Finastra names a partner, Marketnode, as the supplier of the large language model based document automation behind loan onboarding, in a dated announcement with the integration point identified by module, and separately lists third party artificial intelligence applications from named partners in its marketplace. That is genuinely more than the roster norm and is useful to a buyer performing third party risk assessment.

It is still not model disclosure: no base model, provider or version is named for the partner capability or for any of the company's own branded products, which are described only as generative. Recorded as a distinct sub shape of the standing rule, alongside naming the data source and naming the cloud: naming the partner is not naming the model, though it is the most informative of the three.

Graded A on the basis that carried Azentio, SBS and Kiya.ai, and more strongly than any of them for lending specifically: the vendor supplies the system of record for the loan itself, at a scale where it is the de facto market infrastructure for syndicated lending.

Loan IQ holds a single data model across the whole commercial portfolio, and the company sells a dedicated market integration layer, Loan IQ Nexus, whose stated purpose is connectivity and interoperability across both internal and external market platforms, which is an integration product rather than an integration claim.

Around it sit a universal banking core, a payments hub, financial messaging, open application programming interfaces, and a partner application marketplace with third party products listed against the lending line. The documentation product is described as integrating with most core banking platforms and third party data sources. Deployment supports on premises and container based cloud with a named managed hosting service.

Deployment shape is answered and residency is not. The lending platform is stated to support both on premises and container based cloud deployment according to client need, and a named managed hosting service is sold alongside it, with the documentation product separately offered in a cloud edition. That is a real architectural statement rather than a generic cloud claim.

Against it, nothing names a cloud provider, a region, a data residency position or a transfer mechanism, for a vendor holding loan and borrower data across a global customer base including institutions inside the European Union, the United Kingdom, the United States and Asia. Graded C on the same basis as Loxon, Pennant and Kiya.ai, all of which publish deployment choice and no residency.

No pricing published for any product in the estate. No tiers, no bands, no licensing model, no indication of how a syndicated lending platform, a documentation product sold to community institutions and a managed hosting service are each charged, despite those being three very different commercial shapes sold to buyers three orders of magnitude apart in size. Every route ends at a contact form. A software directory carries a pricing entry for the lending platform and it is disregarded here under the standing rule that commercial transparency is never graded from an aggregator.

The broadest coverage on this roster alongside SBS, and unusually it is published as an explicit segmentation rather than inferred from a logo wall. Nine institution segments are addressed by name: community banks, credit unions, small business banks, mid tier financial institutions, high tier banks and financial institutions, corporate and commercial banks, non bank financial institutions, corporates and fintechs.

The lending footprint spans the full size range in a way few vendors can claim, from more than three thousand community, regional and national institutions on the documentation product at one end to twenty one of the top twenty five syndicated lenders and nine of the top ten global agent banks at the other. Buyer types extend past banks to third party asset servicers and private credit lenders, and asset coverage runs syndicated, bilateral, small business, commercial real estate and export finance.