Skip to content

Augmented intelligence:
A pragmatic look at AI in trade

As AI brings new opportunities to global trade, we explore whether it can drive large-scale transformation and if so, in what context.

8 October 2025

8 mins

Graphic representation of microchip with letters AI

Despite decades of technological advancement, paper continues to be a big part of global trade: about 25 billion physical documents are estimated to be processed daily to support just one aspect of global trade – container shipping. That’s not due to a lack of technology; it’s due to structural complexities that previous advances have struggled to fix.

Today, artificial intelligence (AI) brings new opportunities to transform trade. What remains to be seen is whether AI can drive large-scale trade transformation that has eluded earlier technologies, and if so, in what context?

Samuel Mathew, Global Head, Documentary Trade at Standard Chartered, believes that it could be a possibility, but certainly not in the shorter term.

Taking a pragmatic approach to AI’s potential in revolutionising trade is essential. Both corporates and financial institutions need to learn from the challenges faced by previous technologies, such as blockchain, and ensure that AI is set up for success to transform trade.
Samuel Mathew
Global Head, Documentary Trade, Standard Chartered

Why scaling trade digitalisation is complex

Understanding AI’s potential impact on global trade, and documentary trade in particular, starts by grasping the extent of paper-based documentation in trade, which underlines the challenge of disrupting the status quo.

One of the leading factors that hinders digital transformation at scale and the reduction of paper in global trade is interoperability. In the early days of trade digitisation leveraging blockchain for e.g., the challenge was around the costs involved in deploying on-premises nodes conforming to data and cloud regulations. Thereafter, the challenge particularly for banks, their clients, and counterparties, was to join a wide range of digital business networks and platforms that proliferated across the ecosystem in turn hampering stakeholders’ ability to interoperate and scale.

Fintech trade platforms tried to address this problem; but because – to a larger extent – they were not open networks and their revenue models were tied to exclusivity, this led to the rise of “closed digital islands,” stymying the flow of electronic documents across the system. Additionally, businesses were unwilling to share trade data across open networks or lacked incentives to connect and share data, while the platforms’ closed operating models created a friction-filled ecosystem.

This meant insufficient adoption for any single platform, which then tended to be sub-scale. Now, there are organisations and initiatives focused on global trade data standards (such as ICC-DSI, DCSA, BIMCO, SWIFT) that serve as interoperability enablers by establishing digital standards and leading pilot interoperability projects. Whether these scale in the digital trade platforms space remains to be seen.

“Interoperability was the biggest hurdle. The fragmented landscape meant no single platform was adopted by buyers, sellers, carriers and banks spread across different jurisdictions, while key stakeholders couldn’t identify practical, scalable use cases,” says Mathew. “Unless all parties are part of the same platform, the system doesn’t really flow or scale or drive efficiencies, so you end up going back to paper”.

Another core challenge is the actual incentive to move away from paper. While some trade finance, particularly in the open account space – which accounts for more than 85 per cent of global trade – is digital, the underlying physical trade involving the movement of goods across borders still relies overwhelmingly on paper.

UNESCAP estimates 95 per cent of bills of lading documents are in physical form, a figure that underlines the extent of the digitalisation challenge (and the opportunity thereof). And a lot of those documents flow via banks, which perform the role of a trusted intermediary for risk mitigation, financing and settlements.

It is crucial here to distinguish between the digitisation of the underlying physical trade between a buyer and seller, and that of the trade finance – in which lenders such as banks are involved. Not all trade involves financing, and not every trade transaction requires a bank. The real challenge lies in digitising at scale the underlying trade between the buyer, seller, and relevant logistical chain; once this foundation is in place, trade finance will naturally follow.

Given these core issues, paper is likely to feature in global trade over the foreseeable future. This is why Mathew argues for the need to focus on AI’s augmentative abilities in the shorter term in order to truly realise its full potential.

AI’s most impactful application in the short-term lies not in eliminating paper, but in intelligently enhancing companies’ abilities to process documents more efficiently and helping teams perform generic tasks better,” he says. “The goal should not be blind automation, but intelligent augmentation.
Samuel Mathew
Global Head, Documentary Trade, Standard Chartered

The importance of identifying the right trade finance use-cases

A critical starting point for financial institutions and particularly banks, when setting their trade AI strategies and roadmaps, is identifying the business problems they need solved clearly, while considering whether that solution can be quickly scaled up to ensure success.

“Get the trade use case right, answer the scalability question upfront and select the right AI tools from the multitude available, ranging from predictive large language models to supervised and unsupervised learning models,” Mathew advises. “If you get the task and the tool right, you will be successful in addressing the scale challenge.”

In line with this, there are several core trade use-cases today where AI can have significant impact in global trade over the short term, which banks are already working on such as:

Automated document handling

AI tools are getting better at extracting and checking data from various trade documents, far exceeding the performance of past optical character recognition (OCR) tools. This could deliver a phased transition to more automation and boost efficiencies in paper-based workflows – for instance, by leveraging large language models (LLMs) to review and analyse trade documents; accurately extract and digitise information from documents like bills of lading and invoices; and automate the process of comparing data extracted from various trade documents with the specific terms and conditions stipulated in a letter of credit.

Better risk management

AI models could analyse datasets to better assess the creditworthiness of counterparties and predict default risk; and combat financial crime by flagging fraudulent patterns in real time, spotting incidences of double-financing, and indicators of trade-based money laundering like suspicious invoice mismatches, unusual shipping routes or phantom shipments.

Streamlined client onboarding

Agentic AI models could augment or automate bank-wide processes, including Know Your Customer (KYC) and Customer Due Diligence (CDD), by verifying corporate documents like financial statements and board resolutions. Banks and corporates can also enhance the customer experience with AI-powered agents using audio, video and online channels.

For now, though, each of these use-cases will still require a “human-in-the-loop” to address AI’s tendency to hallucinate. Human oversight provides the guardrails that ensure the ethical and responsible use of AI, especially when high-stakes business decisions are involved. This won’t change until AI models fully mature.

Maximising AI’s impact through strategic enablement

Beyond the importance of use-case identification, tool selection, and scalability, there are four essential enablers that both corporates and financial institutions should address to maximise AI’s transformative potential in trade.

Infrastructure and data

It is essential to budget for the cost of overhauling legacy systems without which sophisticated AI models won’t deliver benefits. This is also vital to ensure the creation of clean, unified datasets to train the AI models, because these tools are only as good as their underlying data. Yet another consideration that impacts budgets, especially for institutions in highly regulated industries, such as banks, is contending with compliance and regulatory requirements as these put cost-effective SaaS models out of reach and necessitate more expensive on-premises/private cloud solutions.

Cross-functional collaboration

Successful AI integration requires breaking down solos. Organisations must ensure that product and operations teams work together closely with AI engineers, scientists, and compliance experts. This ensures the technical possibilities are grounded in strategic business needs and regulatory requirements, and ensure the right use cases are selected, the correct tools deployed, and adoption challenges anticipated early.

People

Organisations should invest in transforming their workforce by training their staff to use AI tools confidently and strategically, from frontline employees to senior decision-makers. This includes fostering a culture of experimentation, encouraging teams to challenge outputs, and building learnings to tackle hallucinations or errors. Without talent development, even the most advanced AI models risk being underutilised or applied incorrectly.

Guardrails and responsible AI

Beyond performance, organisations must Prioritise responsibility and establish a robust AI governance framework that embeds accountability, transparency, fairness and ethical usage of data. This not only helps reduce regulatory and reputational risks, but also builds long-term trust with clients and regulators.

Looking ahead: AI’s promising role

While previous attempts at scaling trade transformation – such as through blockchain or platform-led initiatives – have been slower to bear fruit, AI adoption unlocks the opportunity for faster treasury transformation. This is because its solutions are not necessarily reliant on platforms or networks, and the high-impact AI use cases are primarily focused on augmenting and automating human tasks, driving efficiencies, and improving client experience.

For AI to drive large-scale transformation in global trade, the conversation should move beyond the hype, focusing on strategically solving for the right problems.

“The next three to five years will be defined by intelligent augmentation in trade  – enhancing human capabilities to drive a new wave of efficiency across both banks and corporates,” notes Mathew. “This isn’t a sudden leap to full automation, but a journey where our people and the AI models mature together, paving the way for selective, low-risk automation. This strategic evolution is the critical path to the future we envision: a world of seamless trade.”

Explore more insights