Key figures from the banking and fintech sectors convened in New York in May for a roundtable discussion on the promises and pitfalls of digitising trade finance. With AI breakthroughs already reshaping the landscape and groundbreaking bank-fintech collaborations in progress, the discussion turned to overcoming industry inertia and seizing transformative opportunities.
Roundtable participants:
- Ben Arber, CEO, Complidata
- Geoff Brady, global head of trade and supply chain finance, Bank of America (host and chair)
- Alisa DiCaprio, chief economist, R3
- Bayo Gbowu, director, trade & working capital sales, North America and global sector lead healthcare, consumer & wellness, Citi
- Mariya George, co-founder and CEO, Cleareye.ai
- Jordane Rollin, managing director, head of trade and working capital, Americas, Standard Chartered
- Juanjo Ruiz, product manager, IBM Connected Trade, IBM
Brady: I think most people agree that the promise of digitisation in trade has been evident for years, but progress has lagged. There are several reasons for this, which we will discuss. Trade finance is an excellent use case for digitisation due to the need for document exchange, cross-border payments, and other processes that benefit from digital solutions.
One topic of interest is how banks, fintechs and large software companies have been collaborating in this space. What kind of developments have you seen on this front, and is it effective? For instance, are we leveraging supply chain data for real-time financing and achieving a more digital landscape for document and payment transfers? What’s your view?
Arber: I agree that trade finance is a fantastic use case for digitisation, but progress hasn’t been as extensive as we’d like. Creating industry standards is challenging because it involves transforming the entire ecosystem, from shippers and exporters to banks, and all the related documents.
Despite these challenges, good progress is being made in individual parts of the ecosystem. Collaboration between banks and fintechs is more positive now than ever. Microservices in various pockets of the industry allow for quick testing of smaller aspects of digital solutions without needing extensive transformations or long onboarding processes.
For me, one of the most exciting features of this collaboration is the use of AI. Machines can now identify anti-money laundering (AML) red flags and check documents under letters of credit (LCs) with high accuracy – tasks we once thought only humans could do. This progress is a result of fintechs’ expertise combined with banks’ willingness to innovate.
George: In the past five to 10 years, collaboration between banks and fintechs has really increased. In the retail and consumer space, using some kind of fintech product to manage finances is now common.
This shift has accelerated in trade finance in the last couple of years. Previously, the technology wasn’t ready, but recent advancements in AI and large language models have made great strides. Trade finance is now poised for disruption, which can be best achieved through bank and fintech collaboration. Fintechs excel at focusing on specific problems, making them ideal partners for banks in this transformation.
Brady: Following up on Ben’s point about addressing small parts of the system rather than tackling the entire structure at once: how receptive have banks been to this approach? Historically, banks would rely on a single platform provider to handle everything. Are they now more willing to adopt an ‘à la carte’ strategy, choosing specific solutions that fit their needs? Are they prepared for this from a technology and strategy standpoint?
Arber: The industry has moved considerably in that direction. Banks are comfortable using sandboxes to test multiple supply chain providers or automation solutions to find the best fit.
Data security remains a challenge, but interestingly many banks are now able to share data – including customer names – under NDA instead of requiring full master service agreements (MSAs) or onboarding. Even when MSAs are needed, banks can now negotiate them in a few months rather than years.
We’re seeing more of an experimental approach, where banks identify specific, well-defined problems and then look to multiple fintechs to solve them.
For example, we’ve had clients ask to automate the extraction of fields 46a and 47a in an LC. It’s a very narrow use case and a small part of the process – it’s not even the whole document – but it’s clear and specific. Banks want quick, effective solutions for these targeted issues, and are looking to fintechs to fix them.
DiCaprio: We should also ask ourselves the question: what is the goal of digitisation? Is it financial inclusion? Or client efficiency? Why are we doing it?
Gbowu: I can definitely attest to the fact that our clients are looking for efficiency; they’re looking for the best way forward. And if the best way forward is through digitisation – or through a fintech – so be it. Historically, banks have been reticent, but we’re now seeing increased collaboration. Clients want the globality and the robust balance sheet of a bank combined with the sleek, user-friendly platforms of a fintech.
We hear from clients about their interest in e-commerce and how platforms like Amazon and Shopify can help. At Citi, we’re focusing on this trend and the bank recently invested in Hokodo, a provider of b2b buy now pay later and digital trade credit solutions. This kind of collaboration is especially crucial for small to medium-sized businesses. Helping this segment of the industry strengthens the overall ecosystem.
DiCaprio: So the objective is not to improve profits necessarily, it’s to respond to client needs?
Gbowu: At the end of the day, I think we’re all client service managers in finance or tech. While there’s always a revenue goal, mastering client service helps us achieve that. If we excel in serving clients, we can effectively generate revenue.
Brady: Banks tend to start with what clients want. Ben highlighted that progress comes from identifying specific problems that can be solved with digital solutions, rather than just aiming for general efficiency. By asking clients focused questions about their issues, banks can better align their efforts and address concrete needs, rather than simply promoting digitisation for its own sake. This targeted approach helps us serve clients more effectively.
Arber: Another big factor driving change is the role of people. There’s a focus on using AI to enhance decision-making and address challenges such as attrition and retirement among experienced and knowledgeable trade finance professionals. Rather than the conversation being about replacing human jobs with automation, the emphasis is on empowering individuals through training and leveraging AI to facilitate knowledge transfer and skill development.
Rollin: My take is that there are two key drivers of digitisation: efficiency in client processes and servicing, including real-time reporting and streamlined interfaces, and the utilisation of data. The fragmented nature of data in trade finance means digitisation not only improves efficiency but also enhances data transparency, aiding in fraud management and credit modelling and assessment – as well as being able to push the envelope in terms of client support because you have a better view on their balance sheet and payment patterns, for example.
To your point Ben, knowledge in trade finance is also very fragmented. Advancements in natural language processing (NLP) have revolutionised knowledge management in the industry, facilitating talent retention and training.
Digitisation is about more than just converting physical documents into electronic formats. It’s about transforming unstructured data into structured formats. This step is crucial for enabling various processes like AML checks and achieving better client servicing. We mentioned fields 46a and 47a; that’s completely unstructured data and a real challenge. We need a structured format to open up possibilities for leveraging data in multiple ways.
Ruiz: Initially, banks either partnered with fintechs or developed their own systems. Later, some fintechs began competing directly with banks, challenging their business models.
Now, with abundant innovation, the focus is on addressing key pain points by partnering with the relevant service providers. IBM, for instance, aims to provide a way for banks to leverage innovation securely and at scale. This involves building robust interfaces for data exchange across various players and ensuring data security and compliance, which is crucial in the global trade industry with all its regulations. So, while converting unstructured data to structured data is essential, establishing a connected trade gateway that facilitates the flow of that data is equally vital for driving the digital journey.
Brady: Are banks showing increased openness to adopting managed service environments to aggregate solutions from fintechs?
George: Some large banks are open to adopting managed service environments, but regional and super-regional banks are still in the process of recognising the need for it. While they may not be fully open to it yet, they acknowledge the necessity and are moving towards realising its importance. I don’t think these banks will be able to do some of the things we’re talking about without such environments in place.
Brady: Is a managed services environment the optimal solution for consolidating various intelligences into one platform in the trade finance space, rather than engaging with individual software providers separately?
Ruiz: To Mariya’s point, different banks have diverse needs based on their size and resources. Budget constraints often drive decision-making, leading banks to start small and focus on specific areas where improvements can be made in cost optimisation or revenue generation to better serve customers. Beginning with one or two software providers to address specific needs allows banks to demonstrate success before expanding into other areas. Managed service providers play a crucial role in managing relationships between banks and software vendors, ensuring cohesion and mitigating risks throughout the transformation journey. IBM emphasises the importance of data security, compliance and risk mitigation to ensure that the solutions serve their intended purpose effectively.
DiCaprio: I have a different answer. Clients have long requested managed services, indicating a strong demand in the industry. However, funnelling that data for independent analysis is challenging, particularly when it comes to regulatory compliance with things like permissions and privacy. While the demand for managed services is evident, achieving seamless data integration with banking systems remains a hurdle.
Gbowu: The crux of the issue lies in managing data securely once it’s out there. Data can move freely, sometimes beyond intended parameters, posing challenges in maintaining safety and security. It’s crucial to invest in ensuring data security to meet client expectations, regulatory requirements and industry standards. While digitisation offers many benefits, safeguarding data is paramount.
Arber: There’s so much potential value in being able to pull, share and compare data. However, most banks are still cautious about sharing data beyond a single party due to concerns about confidentiality. Despite advancements like confidential cloud computing, which allows for secure data comparison, the industry has not yet fully embraced data sharing. There is huge potential but we’re not there yet.
To the point about selecting the best of many solutions, using APIs it’s easier now than it has ever been to integrate multiple solutions so that banks can combine various processes pretty seamlessly. While global banks may already be leveraging this approach, regional banks are still navigating the process of consolidating solutions or utilising APIs to address their needs.
Brady: Is there a technology solution to address the risk control and legal issues banks face with data sharing? Is that something the industry can collaborate on? What’s the path forward?
DiCaprio: Over the years, we’ve seen collaborative efforts to share data like KYC registries, but these haven’t scaled. The biggest advances in moving data have been internal, such as tokens that move value between branches of the same bank.
An indirect issue that your question raises is that data sharing is constrained by regulations intended to control risk as a way to maintain stability of the financial system. But the internet-based technologies we’re talking about now, like AI and DLT, introduce novel ways of collecting, analysing and sharing data that don’t match with traditional categories of risk.
Could we accept a financial system that embraces rather than prohibits these new forms of risk? That is the path forward in this scenario. Regulators recognise this but have been hesitant to embrace this shift. A side effect of this is that it constrains banks to the activity of exploration of new technological solutions without scalable implementation at the global level.
George: Using AI models, there is a way to share the model learnings without sharing the data itself. Each bank can maintain its data within a private cloud while sharing insights and learnings across institutions, enabling advancements without sacrificing data security.
Rollin: That’s a good point, but it’s relatively recent. Historically, you would have to share or host data with a managed service provider, which included an element of regulatory constraint. Given the complexity of trade finance across multiple geographies and regulatory regimes, this was a challenging approach for banks to adopt.
Now, with advancements like processing LCs using language models, banks can train models on their own data and share insights without sharing raw data. However, this approach is still evolving, requiring efforts to distinguish between data and insights and address IP and copyright issues. The industry is still adapting to this shift.
Ruiz: There has been a lot of progress in recent years, particularly with technologies like IBM’s watsonx.data, which enables data to reside in a specific location and be leveraged for analysis, with the ability to share the outcomes externally. Given their heavy regulation and data protection requirements, banks can ensure security and compliance while accessing necessary data for calculations and receiving outcomes in return.
Brady: Given we are a US-focused group but with an international flavour, what are your thoughts on how the US stacks up with the rest of the world in terms of its digital-friendly legislative environment in the trade space?
DiCaprio: It’s improving! An excellent example is the current Uniform Commercial Code (UCC).
In 2018, it had no applicability to digital assets. In 2022, the Uniform Law Commission (ULC) developed a new model law which included extensive amendments, including the introduction of new article 12 in order to cover digital assets. The challenge now is that while the ULC recommends the adoption, it is up to individual US states to enact it.
To date, only a few states have adopted it and, crucially, New York has not.
While there is progress, the US lags behind the rest of the world due to its federal system and slow-moving regulatory architecture. Countries moving fastest in digitising trade, like Singapore, the UK and the UAE, have dedicated officials in government responsible for this area.
Rollin: It’s harder to align regulations across many states compared to a single state. For example, Singapore’s smaller, trade-focused economy and manageable regulatory environment allow it to push forward with innovation more effectively, creating a conducive ecosystem for progress.
Gbowu: From the US standpoint, we can do a lot better. Currently, people are using ‘phantom documents’, to perform aspects of digital trade; they’re essentially documents that move things forward but hold no legal weight. We need to transition from physical to digital documentation that holds legal weight and strong legal backing. Jurisdictions like the UK, Singapore, the Middle East, and even India are outpacing the US in setting efficient parameters and guidelines. Hopefully, New York will soon opine on the matter to provide some impetus for the rest of the country.
Arber: The Electronic Trade Documents Act (ETDA) in the UK is seen as a potential best practice model and is starting to make a difference in day-to-day operations. The act was in part driven by UK regulator action, such as the 2019 ‘dear CEO’ letter on trade finance, but also by the dynamism of a group of industry experts looking to make positive change.
In Asia, regulators and multilaterals like the Asian Development Bank are also very focused on digitising trade due to their export-heavy economies, setting them apart from the US.
In the US, regulators tend to remain cautious about banks using AI and new technology, especially where it touches compliance and financial crime risk. While US regulatory bodies like the Financial Crimes Enforcement Network are interested in promoting new technology and innovation, this is not necessarily well-known within the banking community. This results in reluctance among US banks to use AI in trade finance compliance. In contrast, regulators in other countries, such as France, are actively using AI to examine banks and identify risks that manual systems might miss. US regulators are likely still some time away from adopting such practices.
Brady: What economic benefits and value are we missing out on by not accelerating digitisation efforts?
DiCaprio: The original driver of digitalisation was the global financial crisis. The goal back then was to support compliance with the cascade of new complex regulations. More recently, Covid accelerated digitalisation because physical documents became difficult to move. This suggests that banks need more than the promise of economic value to accelerate digitisation. This goes back to the earlier question of what is the objective of digitalisation.
Brady: What do you think the goal could be from a big-picture economic value point of view?
DiCaprio: There are obvious economic benefits to a fully digital workflow. If we had that, we could offer natively digital products and services that are not simply replicas of existing operations which grew out of paper-based workflows.
This expands the size of the market.
A second benefit is that, as we discussed, it reduces frictions for existing clients. They benefit with lower costs and higher profits which are likely to contribute to economic growth.
Ruiz: I think at the company level, reducing your cost to serve allows you also to serve the underserved.
DiCaprio: But we see almost no evidence of that in trade finance.
Gbowu: You mentioned underserved communities. Digitisation has brought a level of financial services to people who might not have access otherwise, especially in remote areas. This is evident in working capital and supply chain finance, reaching deep-tier suppliers who need assistance and fair access to capital. From a microfinance perspective, and drawing from my Nigerian background, digitisation has enabled people in West Africa to access financial apps and trade finance solutions they never had before. This is directly related to digitisation; it has impacted microfinance and the broader financial ecosystem.
Rollin: The trade finance gap, now at US$2.5tn, keeps growing. Digitisation alone won’t solve everything, but it offers significant opportunities for financial inclusion. One benefit of digitisation is deep-tier financing, which allows us to assess the risk of second and third-level suppliers – something previously impossible due to lack of data. Now, the financial equation can sometimes actually make sense, even for banks facing their own capital adequacy and return constraints.
Similarly, pre-shipment financing is another area that banks have long tried to address. Collaborating with external parties that handle data enables banks to serve parts of this market alongside alternative finance providers. This collaboration allows banks to expand their services without bearing all the risk alone. The technology supporting these advancements is relatively new and helps banks reach a broader customer base. However, digitisation isn’t a silver bullet; numerous regulations and constraints still limit how effectively we can address the trade finance gap.
Arber: I think it’s interesting to identify which exporters are negatively affected by the trade finance gap. Finding these companies, assessing their credit risk, scoring their financial crime risks and enabling liquidity together is a real opportunity. Technology and digital systems, including the legal entity identifier (LEI) and NLP, are becoming more useful for these tasks. They help identify the underlying exporter and exclude bad actors like shell companies or sanctioned parties. Although risks are increasing, the ability to identify and manage these issues will only improve over time.
Brady: What’s the most promising technology in the trade space today in your view?
George: In the trade finance world, we’ve been trying to solve issues for decades, and recent technology breakthroughs in GenAI and large language models are making a difference. The only way to truly achieve digitisation is by understanding complex, unstructured data. These technological advances can help us do that and will be the foundation for further advancements.
Ruiz: AI more broadly, including GenAI, plays a crucial role in extracting and understanding data more accurately. Its ability to comprehend human communication and present information in a user-friendly way is transformative. One important use case is improving customer experience: customers can simply use instant messaging tools to ask questions. Thanks to integration via APIs and microservices architecture, the system retrieves the necessary information and responds in natural language. AI can be deployed to solve specific problems effectively, unlike blockchain, which relies on network effects. This specificity makes AI particularly promising for current and future applications.
Rollin: It’s really hard to bet on any technology due to the constant buzzwords and scepticism. In the 1990s, there were articles claiming the internet would never take off. Technologies like APIs and digital ledgers require multiple stakeholders to move simultaneously, making widespread adoption challenging. However, with AI language models, you can share learnings without sharing data, allowing collaboration on your own model in a controlled environment. This dual capability of AI in both front-end and back-end processes provides many benefits.
For client servicing, the ability to handle multiple languages is no longer a unique advantage. AI can provide real-time information, eliminating the need for calls or website visits. This extends to client onboarding, where AI can quickly identify missing information in documents, speeding up the process. On the back end, AI can process non-digitised documents, accelerating workflows without needing everyone to adopt the technology simultaneously. This dual functionality and shorter return on investment make AI a game changer in the industry.
Brady: I agree. My view is that the predictive power of AI is incredibly promising. Beyond just trade finance, AI could help banks and investors make better credit decisions. There’s a whole community of asset managers and private equity firms that would benefit from using data and AI to help make more informed credit decisions. Additionally, AI could potentially improve risk assessment. Ideally, banks and regulators could use the same models, reducing the disparity between regulatory expectations and bank actions. AI can help identify the most predictive data, leading to more aligned and accurate decision-making processes.
