The Parisian startup MARBLE announces a Series A of 6.5 million euros led by SMARTFIN, with ADNEXUS, PASSION CAPITAL and 42CAPITAL. It is developing an open source fraud detection and AML-CFT compliance platform, intended to replace the slow configuration cycles of historical tools.
In a financial institution, risk rarely evolves at the pace of the system that must detect it. A new fraud appears, a payment channel becomes more widespread, a regulatory obligation changes, and it is then necessary to modify a rule, adjust a threshold, connect a new data source, sometimes reconfigure an entire investigation chain. On paper, this is a parameter, when in reality for banks, fintechs and payment institutions, the subject can quickly become an IT project lasting several months.
This is where MARBLE comes in. The startup, founded by ARNAUD SCHWARTZ and PASCAL DELANGE, announces a fundraising of 6.5 million euros in Series A. The round is led by SMARTFIN, with the participation of ADNEXUS, the vehicle dedicated to digital trust and regulated professions. Historical investors PASSION CAPITAL and 42CAPITAL are also strengthening their position. The company’s total funding now reaches 9 million euros.
The challenge is not just to add artificial intelligence to a profession already saturated with technological promises. MARBLE wants to install a compliance infrastructure closer to the practices of product and engineering teams: configurable, traceable, connected to internal data and capable of evolving without relaunching a transformation program for each change in risk.
Compliance can no longer work in annual version
The fight against money laundering and the financing of terrorism (AML-CFT) consists of knowing your customers, monitoring transactions, checking lists of sanctions or politically exposed persons, detecting abnormal behavior, investigating alerts and then, if necessary, making the necessary declarations.
The problem is that these operations have long been equipped with large software packages designed to respond to a logic of control, rather than a logic of continuous adaptation. They often offer very broad functional coverage, but also rely on complex settings, long integrations and significant dependence on IT teams, integrators or publishers.
However, financial flows have changed in nature: transactions are now instantaneous, international, multi-channel. Payment institutions, neobanks, crypto-asset platforms and fintechs have accelerated the pace, and fraud schemes do not require authorization from the project committee to evolve. Embedded finance infrastructures must also integrate KYC, anti-fraud and regulatory supervision on a large scale.
According to MARBLE, a significant part of the resources of regulated institutions is today absorbed by compliance and manual processing of alerts. In many organizations, analysts still spend a considerable amount of their time ruling out false positives, searching for scattered information, and piecing together cases.
A platform that wants to bring together compliance, data and product
MARBLE offers a real-time decision engine to monitor transactions, filter AML risks, manage investigations and modify detection rules. Its particularity lies in its open source model and a design primarily designed for developers.
The idea is to allow institutions to maintain control of the system, namely connecting their own data, integrating their KYC tools or their sanctions list providers, writing or adapting detection scenarios, then deploying them with a level of traceability compatible with compliance requirements.
This positioning also responds to a concern that has become central in finance: where does the data reside and who controls the models that use it? MARBLE offers deployment on the client’s infrastructure. A bank can therefore, depending on the company, keep data related to its transactions and investigations in its own environment, and use its own AI models or locally hosted models.
The market is between historic suites and new specialists
MARBLE arrives in a furnished landscape. Large banks are often equipped with solutions such as NICE ACTIMIZE, ORACLE Financial Crime and Compliance Management, SAS, FICO, BAE SYSTEMS NetReveal or FISERV. These players have considerable references, teams and functional catalogs; the downside is that they may have become too cumbersome for organizations that want to modify their risk scenarios more quickly.
Opposing them, new generation specialists have developed, such as COMPLYADVANTAGE, NAPIER AI, HAWK AI, THETARAY or LUCINITY. They sell cloud, automation, risk data and advanced analytics capabilities. HAWK AI, in particular, constitutes a serious European comparable in transactional surveillance and explainable AI. In France, TOPOGRAPH illustrates another approach to compliance, focused on the quality and freshness of corporate data.
The border with the fight against fraud has itself become less clear. FEEDZAI, FEATURES SPACE, now in the orbit of VISA, SARDINE or SEON are particularly present in payments and real-time decisions. As soon as a client seeks to bring together fraud, transaction monitoring and investigation, the categories overlap. This development is similar to that of online fraud detection, where platforms must correlate more signals to decide in real time.
MARBLE must therefore not only convince against a publisher: its toughest competitor is often the existing assembly, historical software, a few custom-developed connectors, Excel files that have become essential, a compliance team that knows every exception, and an integrator that has the system memory. In other words, an operational debt that no one really wants to take the risk of getting rid of.
The European deadline of 2027 opens a window
The regulatory timetable reinforces the interest of this approach. The European AMLR regulation, adopted as part of the new European anti-money laundering package, must apply from July 10, 2027. It introduces a body of rules directly applicable in the European Union, while the AMLA, the new dedicated European authority, must strengthen the supervision of the sector.
This deadline will not automatically create unlimited budgets, but will force the actors concerned to verify the solidity of their controls, their policies, their data and their systems. For publishers capable of integrating without imposing a complete overhaul of the information system, the window is real. The subject of compliance has become particularly sensitive since the questions raised by the Wirecard scandal on the supervision of payment players in Europe.
“All regulated businesses today must do more on compliance and anti-fraud, with the same resources. Without automation, the equation is impossible to solve,” estimates SAUMITRA DUBEY, partner at SMARTFIN. For the fund, MARBLE’s interest is precisely to make this automation an operating standard, directly in the client’s technological environment.
ADNEXUS, which is entering the capital, for its part emphasizes the speed of adaptation. “The ability to quickly detect abnormal transactions or transfer orders is a major challenge for financial institutions,” underlines DIDIER ROSSIGNOL, president of the fund. The promise potentially extends beyond finance, to other regulated professions; but it is in banking, payment and fintech that the need is most immediate today.
AI does not exempt you from explaining
MARBLE plans to devote the lifting to four projects: automating the production of rules, improving the processing of alerts and investigations, integrating agents into workflows and strengthening customer deployment capacity.
The subject must, however, be approached with particular rigor. In compliance, AI is not only evaluated on its ability to save time: an alert must be able to be explained, a file reconstructed, a decision justified before an auditor, a regulator or the risk department. A model that reduces processing time but makes reasoning opaque only solves part of the problem. This requirement for traceability is found more widely in AI deployments in regulated sectors.
MARBLE claims to have more than 100 institutions in production, in more than 15 countries, and indicates that nearly 70% of its customers are located outside France. The company also specifies that 70% of them would have chosen its platform to replace an existing solution.
Its objective of exceeding 5 million euros in ARR in 2027 requires transforming this adoption into enterprise contracts, with the support, security, connectors and governance that regulated establishments expect. Open source can facilitate entry but does not replace the trust required by an infrastructure at the heart of financial control.
The real issue is whether financial institutions will finally agree to treat their control rules as a living product, rather than as software that we no longer dare touch once installed.
