For much of the digitalisation journey, international trade has concentrated on finding electronic alternatives to paper, which was entirely understandable. When commerce depended upon documents being originated, created, printed and physically moved between parties, replacing paper with something that could travel electronically offered an obvious improvement.
We have now reached the point where that is no longer an ambitious enough objective.
A PDF can travel across the world in seconds rather than days, but the information contained within it may still have to be extracted when it reaches the next participant. An exporter creates an invoice containing information already held within its systems and sends it to another party, whose systems then need to read or extract that information. The document subsequently moves further through the transaction and much the same exercise can happen again.
We have accelerated the movement of the document without necessarily changing the way the information itself moves through trade.
This is the issue being explored through the Trade Document Object, or TDO, consultation undertaken by Teesside University in partnership with the ICC Digital Standards Initiative (ICC DSI). Rather than asking only how a trade document can become electronic, the work considers whether the familiar document can also carry structured information that other systems can understand and reuse.
Sounding like a relatively modest development, it could, in practice, take us towards a rather different model of digital trade.
Consider a commercial invoice. To the person looking at it, the invoice is a document, identifying the parties and describing the transaction in a format that allows the reader to understand what has been sold and on what basis. Its visual structure helps us recognise the information and understand the relationship between the different elements.
A computer does not necessarily see it in the same way. Provide a system with an ordinary PDF invoice and, unless structured information accompanies it, the system has to interpret what it sees. Technology for doing this has improved enormously. Intelligent document processing and increasingly AI can identify fields and extract information with considerable accuracy.
But there is a more fundamental question. Why are we extracting some of that information at all?
The exporter that created the invoice likely did not begin with an image and then try to discover the buyer, goods or amount from it. That information already existed within its systems before the invoice was generated.
Somewhere in the process, data became a document. Further along the transaction, we then employ technology to turn the document back into data.
The TDO concept gives us an opportunity to reconsider that circular journey.
The consultation is testing whether a trade document can remain recognisable and readable to the people who need it while also making its information available in structured form for technology to consume. Example invoices and packing lists are being tested in several forms, including PDF/A-3 files incorporating XML and JSON data. Importantly, the work is not declaring that a particular technical approach has already been selected. Businesses and other trade participants are being encouraged to test what works within genuine workflows.
That practical approach is important because international trade has seen plenty of technically impressive solutions that have struggled to achieve widespread adoption. The more useful question is not simply whether something can be built, but whether it works when it encounters the fragmented reality of international trade.
With TDO, that means establishing whether information can be created closer to source, carried with the document and then reused by subsequent participants without continually being reconstructed.
If it can, the implications extend considerably further than eliminating some manual data entry.
The attraction of information created closer to source is obvious. If the seller creates an invoice, information taken directly from that creation process should generally provide a better starting point than information subsequently reconstructed from an image of the invoice.
Nevertheless, that does not mean the recipient can simply trust everything it receives.
Structured information is only valuable if the relying party has sufficient confidence in where it came from and whether it remains what the creator originally produced. A perfectly structured data set of uncertain provenances is not necessarily an improvement on an unstructured document.
The TDO consultation therefore goes beyond the mechanics of embedding data. It also considers metadata, organisational identity and verification, whilst recognising that the usefulness of structured information depends upon the confidence that can be placed in it.
For trade finance, this could eventually become more significant than the technical format itself.
Banks do not examine documents simply because trade finance happens to involve documents. They examine them because the information or evidence represented by those documents forms part of a transaction in which the bank has assumed particular obligations.
If some of that information begins to arrive in structured form, the bank's concern will not merely be whether its systems can read it. The bank will need to understand its provenance and determine what reliance can appropriately be placed upon it.
Those questions are not entirely new. A piece of paper has never been reliable simply because somebody printed it, nor does a PDF become trustworthy merely because it looks convincing on a screen. What structured data potentially allows us to do is become more explicit about provenance and integrity rather than allowing the familiarity of a document to provide an impression of assurance.
This becomes particularly interesting when TDO is considered alongside documentary credit processing.
Banks have invested heavily in technology capable of extracting information from documents. Once extracted, information can be compared with credit conditions and with information appearing elsewhere within a presentation. AI offers the prospect of making this progressively more sophisticated as systems become better at understanding context rather than merely comparing fields.
There is considerable value in that development, particularly because any transition from today's documentary environment will take time.
TDO nonetheless encourages us to look beyond improving extraction.
Suppose an invoice reaches a bank as a human-readable document while relevant invoice information is also available as structured data created when that invoice was generated. The bank's system does not necessarily have to begin by interpreting the visual document to discover every piece of information contained within it. It can potentially access the structured information directly and use the visible document where human examination or interpretation remains appropriate.
This then changes the starting point. Much of today's automation attempts to teach machines to perform more efficiently, something that people have historically had to do. TDO potentially allows us to remove some of that activity altogether because the machine no longer has to rediscover information that already existed digitally.
However, it would be wrong to jump from there to the conclusion that documentary examination disappears. A documentary credit contains conditions against which a presentation is examined, and some questions will continue to involve interpretation. Structured information and conventional documents are also likely to coexist for a considerable period.
What could change is the amount of effort devoted to extracting and re-establishing information that was already known before the document was created.
That could allow technology to handle more of the routine processing while practitioners concentrate increasingly upon those areas where experience and judgement provide greater value.
TDO also deserves attention because of the rapid development of AI within trade finance. Much of the current discussion understandably concentrates on what AI can do. Can it understand a document, identify a potential discrepancy or recognise that apparently different descriptions are not necessarily inconsistent?
These are useful questions, but they begin some distance downstream. AI performance is inevitably affected by the information with which the system is working. When an AI tool receives an unstructured PDF, it first has to determine what information the document contains before it can reason about its significance.
Where structured information accompanies the document, part of that interpretative burden can potentially be removed. The system can use structured information where its provenance and integrity are sufficiently established, while AI can be applied where genuine interpretation is required. Rather than asking increasingly sophisticated technology continually to compensate for unstructured information, we can improve the information it receives in the first place.
This could eventually alter the economics of trade finance automation. Investment does not have to concentrate indefinitely upon making extraction more accurate if some information no longer requires extraction.
The distinction is between becoming better at reading the document and reducing the need to read parts of it at all.
There is also a wider standards discussion behind the TDO work. ICC DSI has already undertaken extensive work through its Key Trade Documents and Data Elements initiative to identify and align the information contained across important trade documents. That provides an important foundation because digital interoperability depends upon participants having a common understanding of the information being exchanged.
TDO takes the discussion into the practical movement of that information. Agreement about what a data element means does not automatically enable one business to send it to another in a usable form. The information still has to be created and transmitted in a way that allows another system to recognise it without requiring both organisations to operate within the same technology environment.
This is particularly relevant when considering the history of digital trade platforms.
A platform can create an effective digital environment for its participants, but an international trade transaction rarely takes place entirely within one environment. An exporter may operate its own ERP system while its logistics provider uses something different. The bank has its own infrastructure and public authorities may impose separate requirements.
The practical objective cannot therefore be to make everybody use the same system.
A more achievable ambition is to allow different systems to understand and use the same information.
TDO potentially contributes to that by allowing the information to travel with a familiar trade object rather than requiring every participant to join a common platform before it can participate digitally.
That could make the concept particularly relevant to businesses that do not have the resources or transaction volumes to integrate individually with numerous digital trade environments.
Interoperability has become one of the most frequently used terms in digital trade, sometimes to the point where its practical meaning becomes obscured. TDO gives us a useful way of making it tangible.
An exporter could create an invoice that remains readable in the normal way while also carrying structured information. The exporter would not necessarily need to create a different digital representation for every party subsequently involved in the transaction. Different participants could access the information relevant to their particular role while retaining access to the familiar human-readable document.
They do not need identical systems, but what they do need is sufficient agreement about the information and a reliable means of carrying it between those systems. This shifts the emphasis away from connecting every platform directly with every other platform and towards making the trade information itself portable.
There will inevitably be practical difficulties. International trade operates across jurisdictions with differing requirements and organisations will reach different conclusions about the extent to which they are prepared to rely upon externally created information. Standards will also continue to evolve.
But these are arguments for testing the proposition rather than reasons for avoiding it.
The fact that the current TDO work is a consultation is therefore important, allowing assumptions to encounter real operational experience before they become embedded within a supposedly finished solution.
Trade finance practitioners have a particular reason to participate while the TDO model is still developing. Banks have accumulated decades of experience dealing with the consequences of ambiguous trade information. Documentary credit practice continually encounters differences in terminology and apparently conflicting information, requiring practitioners to understand both what a document says and what the applicable rules require them to do with it. That experience can contribute considerably more to TDO development than simply confirming whether a bank's systems can consume XML or JSON.
There are deeper questions about what information needs to be verified and how the relationship between the visible document and embedded data should operate. Consider what happens if the human-readable invoice and its structured information are inconsistent. That takes us quickly beyond technology.
Is the structured data a representation of the document, or is the document a human-readable representation of the underlying data? Which should a bank examine if they differ? Does the answer change according to the purpose for which the information is being used?
There may eventually be clear answers, but they should emerge from careful consideration of real transactions rather than assumptions embedded within technology. This is precisely why banks and experienced trade practitioners should engage before the model becomes settled.
There is also a longer-term question that should not be avoided. TDO deliberately retains the familiar document because international trade continues to depend heavily upon documents that people can read. That makes the approach practical and could reduce the barrier to adoption. We should nevertheless be careful not to turn a transitional advantage into a permanent architectural assumption.
If structured information becomes increasingly reliable and reusable, some trade processes may eventually require the visible document far less than they do today. Others will continue to need it because the document performs a commercial or legal function that cannot simply be reduced to a collection of data elements. We do not yet know where that boundary will eventually end, and we do not need to know before experimenting with TDO.
One of the attractions of the concept is precisely that it does not require international trade to leap directly from today's documentary environment into an entirely data-based future, because it potentially provides a bridge between them. Businesses can continue producing something familiar to people while progressively making the underlying information more useful to technology. Experience can then determine how much of that bridge we ultimately continue to need.
The success of TDO should ultimately be judged by something more demanding than whether the technology works. Embedding structured information within an electronic document is technically achievable. The more difficult commercial question is whether another participant can use what has been created without introducing a new layer of complexity that outweighs the benefit.
Can an exporter create the object without making its existing processes significantly more difficult? Can it travel through normal trade channels without requiring every participant to join another closed environment? When it reaches a bank, can relevant information enter existing processing systems with sufficient evidence of its provenance and integrity for the bank to make practical use of it?
Those questions take us to the heart of what TDO is trying to establish. If the answers become increasingly positive, the concept could help address one of the persistent inefficiencies in trade digitalisation: information being repeatedly created, converted and reconstructed as it passes from one participant to another.
That would also change the direction of automation. Instead of concentrating indefinitely upon improving our ability to read documents faster, we can begin reducing the amount of information that needs to be rediscovered from documents in the first place.
There will still be documents because international trade will not abandon familiar commercial artefacts overnight. There will continue to be information requiring interpretation, while experienced practitioners will remain necessary where judgement cannot sensibly be reduced to a data comparison.
But the balance can begin to change. The Trade Document Object consultation should therefore be viewed less as an attempt to create another electronic document format and more as an exploration of how trade information itself might travel.
For years we have spoken about paperless trade as though removing the paper were the destination. In reality, paper was the medium through which information and evidence travelled because, for most of the history of international trade, there was no practical alternative.
We now have alternatives.
The more interesting stage of digital trade begins when we stop concentrating solely upon replacing the medium and start making the information itself capable of moving reliably between the people and systems that need it.
TDO may ultimately provide part of that architecture or the consultation may reveal that aspects of the approach need to develop in another direction. Either outcome would be useful.
The purpose of consultation should not be to prove that the original idea was correct. It should be to discover what works when an idea encounters the complexity of actual international trade.
That is why this initiative deserves attention from banks, businesses and practitioners while it is still being shaped.
Source and acknowledgement: This blog draws upon the Trade Document Object consultation undertaken by Teesside University in partnership with the ICC Digital Standards Initiative (ICC DSI) and associated consultation material. The discussion and analysis above reflect the author's interpretation of the potential implications of the initiative for trade finance and digital trade.