smartKYC

The removal test leaves keyword search, which is precisely what the company defines itself against and what supervisors criticise. Multilingual natural language processing machine reads online media content, registries and court records, semantic search finds risk relevant material regardless of the language it was written in, and relevance scoring resolves which hits actually concern the person being screened. Relationship inference surfaces associations with politically exposed and high risk persons. None of that is achievable with lists and string matching.

The design is exception based and the institution's own policy governs it: events are classified by criticality so that alerts and their escalation pathways reflect client risk policy, and continuous rescreening runs in the background alerting the analyst only by exception, which is the correct shape for monitoring at a volume no team could read.

Straight through processing is offered for onboarding, watchlist filtering and batch remediation, which is higher automation, and the company has published thoughtfully on agentic systems that would decide when to escalate and which sources to prioritise, framing that as promise and precaution rather than as a shipped capability. What is absent is any stated threshold or confidence measure attached to an individual escalation.

Findings are grounded rather than generated, with risk conclusions stated to come from multilingual language processing applied to verifiable source content, so every hit traces to a document a reviewer can read. Explainability is a standing subject in the company's published material rather than a marketing line.

And auditability is treated as a product requirement because it is a supervisory one, with the ability to demonstrate why a hit was discounted or escalated named among the failures examiners find. What is absent is measurement of the system's own performance: no precision, recall or validation result is published, and for a screening engine whose entire claim is fewer false positives at equal or better detection, those are the figures a buyer needs.

Twelve years of operation with no customer named anywhere is the central gap. An executive refers to several clients performing frequent adverse media rescreening, and trade press in the wealth management sector covers the company and quotes it as an authority on the category, which indicates standing in the market it serves.

Independent software directories list and review the platform, noting that multilingual search and federated source coverage deliver stronger screening quality and fewer false positives. No funding, headcount, client count or volume figure is disclosed, so the scale of deployment cannot be established.

No data boundary statement was located. The platform screens against public and licensed sources rather than pooling customer data, which limits the usual concern, and it nonetheless accumulates harmonised entity profiles and the record of which hits each institution discounted or escalated, and that adjudication history is exactly what would improve relevance scoring across the customer base. Nothing states whether it is used that way, whether an institution's screening decisions remain its own, or how entity profiles are separated between banks screening the same individuals.

This vendor engages the privacy problem inherent in its own category rather than ignoring it, publishing specifically on how different legal systems treat privacy and transparency when it comes to naming individuals in media reports, which is the question that determines whether an adverse media hit in one jurisdiction should be actionable in another.

European data protection is a standing subject in its published material, and the company describes its approach as rooted in anonymity and auditability. That is more consideration of the subject's position than any comparable vendor demonstrates. Held at B because no data protection agreement, retention schedule or subprocessor list was located, and the platform necessarily holds risk profiles on named individuals who never dealt with it.

No attestation, certification, trust centre or enumerated framework was located. Twelve years of serving private banks and investment banks means vendor security assessment has been passed many times, and for a platform holding client risk profiles on behalf of institutions whose own confidentiality obligations are among the strictest in banking, publishing that control set would be the natural complement to the regulatory rigour evident elsewhere in its material.

The regulatory grounding is precise at three levels. The international standard setter for anti money laundering is named, the specific obligations are named as customer due diligence and enhanced due diligence, and the national implementing regulations are cited.

More useful still is how the company frames the problem, describing what supervisors consistently find when they review adverse media programmes through thematic reviews, routine examinations and enforcement matters: screening that operates only at onboarding, keyword searches that drown analysts in irrelevant alerts, coverage that stops at English language sources, and audit trails that cannot demonstrate why a hit was discounted or escalated. Four named supervisory failure modes, with the product built to answer each, is the strongest regulatory articulation in this index.

This vendor describes the mechanism rather than the aspiration on the adverse media false positive problem, which almost nothing in this index does. Hits are scored for relevance using other identifying attributes such as age, nationality and company affiliation rather than by name alone, in any language, which directly addresses the harm recorded repeatedly here, that common names and non Western naming conventions generate false matches which then follow innocent people.

Multilingual coverage addresses the second half of the same problem, since screening that stops at English systematically under detects and misreads material about people outside English speaking countries. The company also publishes on what happens when these systems get it wrong and on the jurisdictional privacy question in naming individuals. What is missing is measurement: no false positive rate, per language accuracy or error analysis is published.

No commercial guarantee or indemnity was located, and the auditability commitment is substantive because it is aimed at the right requirement. The company identifies audit trails that cannot demonstrate why a hit was discounted or escalated as a failure supervisors repeatedly find, and builds to answer it, so an institution can reconstruct and defend every adjudication. That protects the bank and indirectly the subject, since a recorded rationale is a reviewable one.

The person screened still has nothing directly: they are profiled without knowledge, and the company's own published analysis of jurisdictional privacy shows it understands that position better than most without offering them a route.

Source categories are enumerated clearly, covering online media, company registries, court databases, social media platforms, registry filings and watchlists, with federated search running across them, so a buyer understands the evidentiary basis of any finding and that conclusions rest on retrievable documents rather than on an opaque risk database. The company also publishes on open source intelligence as a discipline, explaining what it captures that conventional data providers miss. No individual provider is named for any source, no model provider is identified for the language processing, and no subprocessor list or hosting arrangement was located.

One integration is described with genuine specificity and it closes a real loop: suspicious transaction monitoring systems can be connected so that a flagged payment or an unknown beneficiary automatically triggers enhanced due diligence on the parties involved, in real time and at scale, which turns two separate compliance functions into one workflow.

Federated search across multiple source types means the platform reaches registries, court records, media and watchlists without the analyst switching tools. Straight through processing supports onboarding, watchlist filtering and batch remediation. What is not published is any named system, so an institution cannot confirm its own monitoring or onboarding platform is supported.

No hosting provider, region selection, residency commitment or private deployment option was located. The question carries weight for this product because the company itself writes about jurisdictional differences in how personal information may be handled, and it serves private banks across multiple regimes while holding risk profiles on named individuals, so where that processing occurs is a question its own published analysis implies it has considered.

No pricing, packaging or basis of charge was located, and third party software directories list the platform without pricing. One adoption cost signal appears in independent reviews and is unusually candid for coming from outside the company: setup and tuning can take time, which for a configurable screening engine is the material implementation consideration. Nothing indicates whether charge falls per screen, per monitored client, per seat or as a licence.

Five financial institution types are served alongside corporates, spanning private banking, corporate and investment banking, retail banking, wealth management and compliance advisory firms, and multinational companies use the same tooling for third party and supplier screening. Lifecycle coverage is complete rather than episodic, running from historic screening at onboarding through periodic refresh to real time continuous monitoring, with batch remediation for existing books. Language coverage is the differentiating dimension and is global by design. The limit is functional: this is screening and monitoring, and it stops before transaction monitoring or case management.