FIS

The Clearwater precedent, settled by the company's own description of itself. It operates systems that clear payments, move money and run core banking for thousands of institutions worldwide, and every one of those functions predates and survives the removal of any model.

The learned line is real, named, shipped and in one case already deployed at named institutions, which is what makes this a build rather than a rejection: a generative product support tool inside the treasury platform since March 2025, a treasury suite built around it, a revenue insight product, and an agentic financial crimes capability announced May 2026.

What distinguishes this from the platform tier rejections is that the capabilities are separately branded, separately purchasable and attached to named customers, rather than an adjective applied to an existing estate.

The clearest description of an execution path and its stopping point found on this roster. The financial crimes capability is described as assembling evidence across core systems, evaluating activity against known typologies, and surfacing the highest risk cases for investigator review. That is three named steps with the handoff to a person stated explicitly and placed at the decision rather than after it: the agent ranks and presents, the investigator disposes.

Compare the roster norm, which asserts either an outcome or a posture. Held at B and not A on two counts. No threshold, confidence band, default configuration or escalation path is published, and nothing states what the agent may not do, in particular whether it can close or dismiss an alert rather than only escalate one.

And the description appears in dated trade coverage attributed to the company rather than in a product specification, which under the standing evidence test is one step removed from vendor published material.

Nothing. No accuracy or error rate, no precision or recall figure, no validation methodology, no backtesting, no drift monitoring, no versioning and no external assessment, across a product set that includes generative product guidance and a capability that ranks money laundering risk. Evaluating activity against known typologies describes the classification approach and is not validation of it.

The standing question this index now puts to every screening product applies with full force and is unanswered: a capability sold on compressing investigation time must be read alongside what happened to the cases it did not surface, and no recall or false negative figure is published.

Named institutions attached to a named artificial intelligence product, which is better than the logo wall this axis usually sees, and no attributed result. Bank of Montreal and Amalgamated Bank are named among the first deploying the financial crimes capability, with broader availability stated for the second half of 2026.

Independent assessment is strong: category leader across all five quadrants of a 2026 analyst evaluation of credit lending operations, which is a sweep rather than a placement, plus innovation recognition for the treasury product in 2025. Held at B because every number published floats free of every name.

The headline claim is compressing investigations from hours to minutes and it attaches to no institution; the figure for annual United States anti money laundering operating spend is market sizing, not a result. Worth noting for the wider finding on this axis, since the two named institutions are deployers rather than referenceable outcomes.

Silent on every part of the question. Nothing states whether customer data trains or tunes any model, whether anything is pooled across the thousands of institutions on these platforms, what is retained from a generative support session, or what the third party model providers named elsewhere in this record are contractually bound to do with data passed to them.

That last point is the one that matters most here and is the second order question almost nothing in this index answers: naming a model provider tells a buyer who is in the chain, and says nothing about what that provider may keep.

Nothing at product level in the material reviewed. No retention schedule, no deletion terms, no subprocessor disclosure and no tenant separation statement. The unaddressed question is specific and sharpened by the architecture the company itself describes: a capability that assembles evidence across a bank's core systems is by design reading customer transaction, account and relationship data at breadth, and where a third party model provider sits in that path the buyer needs to know what leaves the institution and what is retained by whom. Nothing addresses it.

No certification, attestation report, audit period or trust portal was located in the material reviewed, so the grade follows the standing rule that a credential must be found and named rather than assumed from a vendor's size. One disclosure sits here that is genuinely unusual and is recorded because it is rare rather than because it changes the grade: the company has publicly stated that it applies a frontier model to scan and evaluate its own systems as an added cybersecurity layer, under a controlled access programme aimed at software supporting critical infrastructure, with the reasoning given that it protects code running payment clearing and core banking.

That is a described security engineering practice with a named tool, which almost nothing in this index publishes. It is not third party assurance and this axis measures what a buyer can verify, so it does not lift the grade. Queued and high value, and this is unproven absence rather than evidenced absence: a supplier of this scale to regulated institutions certainly answers security questionnaires and very probably maintains a customer facing compliance portal that was not located here.

Corrected upward on 20 August 2026 after a targeted check, having first been graded C. This firm holds genuine supervisory standing, and it is involuntary rather than sought. Under section 7 of the United States Bank Service Company Act, a provider performing services for a depository institution is subject to regulation and examination by the federal banking agencies to the same extent as if those services were performed by the institution itself.

The Federal Reserve, the Federal Deposit Insurance Corporation and the Comptroller of the Currency hold that authority and coordinate it through the Federal Financial Institutions Examination Council, which runs a dedicated programme for multi regional data processing servicers covering an estimated twelve to twenty of the largest nationwide bank technology vendors, and examined providers are assigned a rating under the uniform rating system for information technology.

A firm running payment clearing, money movement and core banking for thousands of depository institutions falls squarely inside that description, and secondary analysis of securities filings identifies this company among the three largest core providers disclosing the status. Held at B rather than A on two grounds.

The membership list of the supervisory programme is not public, and this company's own securities disclosure was not read directly in this pass, so inclusion is reported rather than verified. And the supervision has limited public enforcement history.

The distinction that matters for the rest of this index is recorded in the sweep notes: this authority attaches to performing the service, not to licensing software an institution runs itself, so it separates processors from licensors rather than lifting every large platform vendor.

No published governance position, no fairness testing, no protected characteristic treatment, no named framework and no statement on any artificial intelligence regulation. The exposure is concrete and under discussed generally, which makes it worth stating plainly: a capability that ranks customers by money laundering risk and directs investigator attention determines who gets investigated, and the burden of a false positive in that process falls on a real account holder in the form of frozen funds, closed accounts and exit decisions, disproportionately on populations already over represented in suspicious activity reporting. Nothing addresses fairness in typology evaluation or risk ranking anywhere.

No recourse position published. The division is the standard one and the consequences here are unusually severe for the individual: the institution files the suspicious activity report, exits the customer and answers to the supervisor, while the supplier assembles the evidence and ranks the risk that led there.

Nothing states whether an affected customer learns a model was involved, how a wrong risk ranking is contested, what happens to a case the model failed to surface, or how responsibility divides between the supplier, the third party model provider named in the chain and the bank operating the capability. The affected person is not the buyer and has no route to any of the three.

The strongest disclosure on this axis found in the entire sweep, and the first grade above C on it from this roster. The company names its model suppliers in its own dated announcements rather than leaving a generative layer unattributed. The treasury support tool is stated to be built with Microsoft and to run on that company's managed model service, which names a provider against a specific named product.

The agentic financial crimes capability is announced as a partnership with a named model developer, again against a specific product with named deploying institutions. And for internal use the company names a specific model version applied to scanning its own systems, which is version level precision almost nothing in this index reaches.

Held at B and not A because the disclosure is partial and uneven: no model or version is named for either customer facing product, there is no model inventory across the estate, no country of processing, and the remaining capabilities including the treasury suite and the revenue insight product name nothing at all. Graded on the same terms any named provider would receive.

Arguably the deepest position on this axis anywhere in the index, and graded A on the same basis as Azentio, SBS, Kiya.ai and Finastra with a further step beyond them: the vendor supplies not only the core banking system of record but the payment clearing and money movement infrastructure around it, plus capital markets and treasury platforms.

The point is made concrete by the design of the newest capability, which is described as assembling evidence across a bank's core systems, an operation only a supplier already sitting inside those systems can perform without an integration project. The lending suite was assessed a category leader in all five quadrants of a 2026 analyst evaluation of credit lending operations, a sweep of the full quadrant set rather than a placement in one.

One named cloud dependency and no residency position. The treasury platform is stated to be migrating to Microsoft Azure, with speed, scalability and security given as the rationale, and the generative support tool runs on that provider's managed model service. Naming the cloud for one platform is a real disclosure and it is the only one found.

Nothing states deployment options across the wider estate, region availability, data residency, or how customer data is kept inside a jurisdiction for a company running core banking and payment clearing across multiple continents and regulatory regimes. Graded C on the same basis as Loxon, Pennant, Kiya.ai and Finastra.

No product pricing published, which is the category norm and is not offset by the company being publicly listed. Segment revenue and growth rates appear in financial reporting and tell a buyer nothing about what a lending platform, a treasury system or a financial crimes capability costs, how any of them is charged, or how a per investigation or per alert model would work for a capability sold on volume compression. Nothing indicative appears anywhere.

Among the widest in the index. The company states it runs payment clearing, money movement and core banking for thousands of institutions worldwide, and its business is organised across banking solutions, capital markets solutions and commercial lending, which spans retail banks, commercial and corporate banks, credit unions, capital markets firms and corporate treasury functions as distinct buyer types with distinct products.

The two institutions named against the newest capability sit at opposite ends of the range, one a large North American banking group and the other a values based commercial bank, which is a useful indication that the same product is being sold across tiers rather than only at the top.