Posted on mongodb google news. Visit mongodb google news

Manual processes in the fintech space are like paying for digital transactions using cash. Because of tight regulation, fintech companies walk a tightrope when it comes to internal processes. However, that doesn’t mean technology cannot lend a helping hand.

M-DAQ Global, a fintech group specialising in forex and cross-border payment solutions, previously wrestled with a series of manual processes that significantly delayed operations. It eventually turned to AI to build an internal automation platform called CheckGPT, powered by MongoDB Atlas. Andrew Marchen, General Manager of Payments at M-DAQ Global, and Wei You Pan, Global Director of Financial Services Industry Solutions at MongoDB, spoke to Frontier Enterprise about the development and impact of the platform.

Navigating roadblocks

To meet anti-money laundering (AML) regulations, M-DAQ traditionally relied on manual background checks, document reviews, and customer onboarding. These tasks were not only time-consuming, but also inefficient, given the unstructured and sensitive nature of the data involved.

As the business scaled, M-DAQ began handling an increasing number of onboarding requests from clients in multiple jurisdictions.

“Adding personnel wasn’t a sustainable solution. Not only did it slow down our response times, but it also introduced inconsistencies due to varying interpretations of internal compliance policies by different team members. This created friction for our customers and posed a risk to our business,” Marchen observed.

The company also struggled to respond to compliance signals in real time due to fragmented data and workflows.

“Much of our AML work, particularly during onboarding, relied on fragmented systems and required extensive manual verification across multiple sources, slowing down decision-making and introducing the risk of human error,” he said.

Because the data arrived in varied formats, extracting meaningful insights often required manual cross-referencing across multiple systems, Marchen added.

Unlocking solutions

To address these challenges, M-DAQ developed an AI-powered know-your-business (KYB) platform called CheckGPT, using MongoDB Atlas as its back end.

“By automating and standardising how we assess and process customer information, CheckGPT enables our customers to access our services more quickly, without compromising compliance integrity,” Marchen explained.

The decision to use MongoDB Atlas was influenced by its flexible document model, which can handle the changing nature of compliance data, including onboarding documents and the rules-based processes they support.

But like any complex deployment, rolling out CheckGPT came with its share of growing pains.

“One of the toughest parts was building a secure system that could handle multiple clients, manage unstructured data, and deliver AI results in real time. We had to make sure each client’s sensitive data stayed completely separate, without slowing things down, and still support quick searches across different data types,” Marchen said.

To resolve this, M-DAQ adopted a database-per-tenant strategy using MongoDB Atlas, giving each client a dedicated environment to prevent data co-mingling and meet regulatory requirements.

“This architecture also made it easier to apply client-specific compliance policies and updates without system-wide disruption,” Marchen added.

According to Wei, supporting M-DAQ’s platform meant addressing two key requirements: strict data isolation for each client and the ability to manage diverse, evolving data formats at scale.

“Because the platform needed to support multiple clients simultaneously, we had to design for strict data isolation to ensure that sensitive customer information didn’t mingle with others. In regulated sectors, that level of separation is critical for compliance,” he explained.

CheckGPT processes a wide range of inputs, from onboarding forms to media reports, which often arrive in unstructured or semi-structured formats. Wei noted that the platform’s schema design made it possible to accommodate new data types over time, which supported ongoing changes to compliance workflows.

Reaping the rewards

Following CheckGPT’s launch, M-DAQ accelerated its onboarding processes, particularly for clients in highly regulated industries. By automating compliance tasks such as customer due diligence, name screening, and media checks, the platform eliminated much of the manual work that previously slowed operations.

Marchen shared the experience of a payments provider that had faced long onboarding times due to manual document handling and frequent false positives during screening.

“With CheckGPT, we now use AI to review customer due diligence documents, score risk levels automatically, and cut down on false alerts. That’s helped us reduce onboarding time by up to 80%, cut false positives by 90%, and let our teams focus on the cases that really matter,” he said.

In the six months following the rollout, M-DAQ was able to onboard 30 times more customers and scale its gross transacted value by a factor of 10, while maintaining compliance standards.

Planning for the future

Looking ahead, M-DAQ plans to evolve CheckGPT into a broader intelligent compliance and onboarding platform that supports secure cross-border transactions. According to Marchen, AI will remain central to this strategy, not only to automate workflows, but also to identify risk signals, adapt to changing regulations, and build trust across its payments ecosystem.

“Now that onboarding is faster and can support more clients, the next step is to use AI for transaction screening and monitoring. This will help us keep up with growth without depending too much on manual checks,” he said.

Marchen also sees opportunities to apply automation more broadly across M-DAQ’s internal operations.

“There are plenty of internal areas where automation can make a difference, from managing workflows and handling exceptions to supporting teams behind the scenes. These improvements help us work more efficiently and focus on what matters to our customers,” he said.

Using MongoDB Atlas, M-DAQ also plans to expand CheckGPT’s capabilities to support more complex, jurisdiction-specific use cases, particularly in markets where compliance rules change quickly.

Wei noted that many fintechs in Asia-Pacific are now building compliance platforms with modular, cloud-based designs that can support real-time onboarding and AML processes.

“One priority is being able to work with different types of data — whether it’s structured, semi-structured, or unstructured. That includes scanned documents, free-text notes, and even audio or video files, all of which play a role in KYC and AML,” he said.

To support these needs, Wei added, fintechs are turning to document-based data models that can adapt to various formats and power analytics across different data types, from text and time series to network relationships and location data.

Machine learning and generative AI are also playing a bigger role. These tools are now being used to build more complete risk profiles, spot issues as they happen, summarise cases, and simplify reporting requirements.

“Architectures that support scale, control where data is stored, and manage who can access what are becoming more common,” Wei said. “They help companies stay on top of evolving regulations and respond more quickly to potential financial crime.”

Article originally posted on mongodb google news. Visit mongodb google news