Adclear

The removal test leaves the manual process the product exists to replace, a compliance officer reading marketing copy against a rulebook. Models interpret creative material across every format a marketing team produces, covering social imagery, video assets, email, articles, paid search and social advertising, websites and live product screens, and assess each against both a firm's own policy rules and the requirements of more than 100 mapped regulatory bodies, then locate the specific failures rather than returning a pass or fail. Nothing about that survives without the models, which is why review time falls from days to seconds rather than merely improving.

The machine flags and locates; people decide and resolve; and the company says so in those terms. Checking combines its own regulatory mapping with the customer's own policy rules, which are described as fully customisable to reflect that firm's interpretation, so the standard applied belongs to the accountable institution rather than to the vendor. Feedback tells a team exactly where to look and invites them to collaborate on resolving it, which is deliberately not an approval decision.

Deployments are described as integrating into a firm's existing approvals process rather than replacing it, and the product's stated purpose is that nothing goes live that should not, making it a gate rather than a generator. Audit trails record the whole exchange.

Three things work in the right direction. Findings are localised rather than scored, telling a team exactly where content falls short and why, which makes each output checkable against the rule it invokes. The loop closes, because the same platform monitors promotions after publication, so a failure that escaped pre approval surfaces rather than staying hidden.

And graduation from a regulator's own innovation programme means the approach has been examined by the body whose rules it interprets. The nearest thing to a quality measure is the first time approval rate rising from around 35 percent to over 90 percent, which describes how well marketing learns rather than how well the model reviews. The number that matters, the rate at which non compliant content passes, is not published.

Exceptional for a company that launched in 2024 with a team of eight. Named customers include three of the largest United Kingdom banking groups, a major life and pensions insurer, the country's leading neobank, two of the world's largest cryptoasset exchanges, and a long list of investment, trading, pension and credit platforms, with two further deployments announced in the weeks before this assessment.

Outcomes are quantified with before and after figures rather than headline percentages: first time approval rates roughly triple from around 35 percent to over 90 percent, review times fall more than 90 percent, cost per review drops 44 percent, and marketing output rises up to eighteenfold because compliance ceases to be the constraint. Volume is stated at more than 50,000 promotions reviewed across clients and over 100,000 assets through the monitoring product, with recurring revenue up tenfold in ten months.

No data boundary statement was located and the question has an unusual shape here. Each deployment begins by importing that customer's policy manual and rule packs, which is per client grounding and points toward containment, but the platform's core value is accumulated regulatory interpretation built from reviewing tens of thousands of promotions across every client, and that pool necessarily includes how competing firms interpret the same rules.

Three of the largest banks in one market sit on the same platform. Nothing states whether one firm's policy positions or reviewed content inform the model applied to another, or whether a customer can decline to contribute.

Structurally favourable by subject matter, which is the Nammu21 and Daloopa property. The material processed is a firm's own marketing content and its internal policy manuals and rule packs, not records about customers, so there is no consumer personal data in the payload and no data subject whose information could be exposed.

What is held instead is commercially sensitive in a different way, since a firm's compliance policy manual describes exactly how it interprets its obligations and where it draws its risk lines. Held at B because no retention schedule, subprocessor list or data processing terms were located for that material.

No attestation, certification, trust centre or enumerated framework was located. Deployments at three of the largest banks in one market and at a major insurer mean vendor security assessment has been passed at demanding standards, and none of that assurance is published, which is the same pattern recorded for Auquan. For a company of this size the absence is understandable and it is also the artifact the next enterprise buyer will request first.

The most extensive regulatory naming in this index by a wide margin. More than 100 regulatory bodies are mapped, and fifteen are named individually in published material, spanning the United Kingdom conduct regulator and its consumer outcomes regime, the advertising code, European markets and cryptoasset regulations, four United States regulators and the advertising authority, and supervisors in Singapore, Australia, South Africa, Cyprus, Luxembourg and Brazil.

The governing standard for financial promotions, that they be fair, clear and not misleading, is named directly. Beyond breadth there is direct regulatory engagement: the company is a graduate of the conduct regulator's own artificial intelligence and innovation programme, and holds a regulatory technology award. Twelfth A on this axis.

No individual is assessed, so the axis adapts, and the governance question becomes what the reviewer fails to catch. A false negative here means non compliant marketing reaching consumers with the institution's approval process apparently satisfied, and that omission is invisible by the same logic recorded for AscentAI: nothing in a clean review signals what was not examined.

Coverage depth is the second dimension, since more than 100 mapped bodies cannot all be modelled to the same standard, and a firm operating under a less thoroughly covered regulator receives a thinner review without being told it is thinner. Post publication monitoring partly answers this by catching what pre approval missed. No false negative rate or per jurisdiction depth disclosure was located.

No guarantee or indemnity was located, and the product nonetheless produces the artifact that matters most when something goes wrong. Every review generates an audit trail described as suitable for internal compliance and regulatory reporting, so a firm challenged on a promotion can evidence what was checked, against which rules, when, and who signed it off, which is precisely what a supervisor asks for.

Post publication monitoring means a problem found later is attributable rather than mysterious. What is absent is anything owed by the vendor when its own review misses something, and no correction or notification process is described for a promotion cleared in error.

The regulatory inputs are inherently transparent, since the rules being applied are published by the authorities themselves and any customer can read them, and the customer's own policy manual is named as the second input. Beyond that nothing is disclosed.

No model provider is named for the interpretation of creative material across text, image and video, no hosting arrangement or subprocessor list appears, and for a product whose function is judging content against regulation, which model performs that judgement is a question a compliance function would reasonably ask.

The platform is designed to sit inside an existing approval workflow rather than beside it, with one named customer integrating it directly into its current approvals process, and separate workspaces isolate affiliates and agencies so external partners can be reviewed without access to internal material.

Input coverage spans every channel a regulated marketer uses, and implementation begins by importing the firm's own policy manual and rule packs, which is the integration that actually determines whether output is usable. What is not published is any named connection to the marketing systems where content is created and stored, so how assets reach the platform is undescribed.

No hosting provider, region selection, residency commitment or private deployment option was located. Operations span the United Kingdom, European Union and United States with further regions imminent, and the material held includes firms' internal compliance policy documentation, which several of the named customers would treat as restricted, so a published residency position would ordinarily be expected.

Everything except the rate is published, which is rare. The charging basis is stated as volume and workflow configuration, so a buyer knows what drives cost. The economics are given as a payback period of under five months alongside a 44 percent reduction in cost per review, which is more useful than a price because it frames the decision the way a finance committee will.

Implementation effort is broken down week by week to a six week pilot, covering policy import, workflow mapping and go live. A tailored return model is offered during evaluation. The only missing element is the number itself.

Seven institution types are served with named customers in each, spanning traditional banks, neobanks, insurers, investment platforms, trading platforms, credit providers and cryptoasset businesses, which is unusually wide for a specialist and reflects that every regulated firm markets.

Regulatory coverage is the second dimension and it is the more remarkable one, with more than 100 bodies mapped and core coverage live across the United Kingdom, the European Union and the United States, and further regions stated as imminent. Channel coverage is complete, reaching email, social, television, radio, paid search, product interfaces and websites, with separate workspaces for affiliates and agencies.