Interview with Mike Rondaij on Trade AI and the future of Trade Finance

One of the most interesting 'new' kids on the block at SIBOS (the annual banking and financial conference ) in London this year is, without a doubt, Dutch based company Conpend.

This Fin-Tech company offers a revolutionary application (Trafinas/Trade AI) which helps financial organizations on their digital transformation journey by empowering their Trade Finance Operations, Compliance-, TBML-, KYC- checks and processes and Relationship Managers to independently increase their productivity and effectivity.

With an appealing and impressive customer base including Commerzbank, ING and UniCredit, we decided to have a talk with one of the creators of the application, Mike Rondaij. With a background of over 35 years in both Trade Finance and Corporate banking, he shares his personal and the company’s view on future of Financial Services in general, and Trade Finance in particular with us.

How did Trade AI come about?

If there was one area that had been passed by the digital revolution it was definitely Trade Finance. Paper based documents and manual labor through and through. Everyone involved in Trade Finance knew this, so it was not particularly unfamiliar to the in-crowd, it was basically a given that automation was presumed too difficult to even look at. So about 5 years ago I was approached by Marc Smith (founder and CEO of Conpend) who asked me if I would be interested to join forces with him and be those daredevils, picking up the glove to change the course of Trade Finance. I said yes and we started, as simple as that.

How did Trade AI obtain such an impressive portfolio so soon after its start?

Well, we did create an amazing product, didn't we? (He laughs out loud] But seriously, without being arrogant, Trade AI is something else, and it's not your average start up product. For starters our team consists of seasoned- and young professionals, and we can rely on solid background in Finance and Engineering both practical and academical. And not unimportant, we are working from within a vision. And it must be said that we owe our customers a lot of credit for trusting and believing in us, because they have been on board from scratch, and in most instances we develop the product and the vision together with them, rather than for them or requested by them. And that, to a great extend, also explains the success. Everyone shares the same view and vision, which is much more a framework and a concept, rather than 'just' a piece of fancy software. And that is basically the force behind the product, we are literally on a mission. We are not just the next, new software company trying to make a quick buck, selling as many licenses as possible. We believe in what we do. That and an awful lot of hard work with a dedicated team of brilliant people.

This year alone we have been in the U.K., Romania, Japan, U.S.A., Singapore, U.A.E., India, Canada, France, Spain, Portugal, Switzerland, Austria, Germany, Denmark, South Africa, the Americas and that list is not even complete yet…For example, I just got back from Italy and next week I’m off to Morocco conducting another PoC.

Can you offer some background to the idea behind Trade AI?

Even as little as 15 years ago, at the dawn of the new millennium, it was still totally unimaginable to think that software would one day be able to digitize, automate or robotize any or all the operational hazards surrounding Trade and Commodity Finance, let alone replace the noble craft of Letters of Credit document checking. In fact, over the last decade the challenge only seems to have increased.

Most of the current challenges we face today surrounding international trade and its administrative settlement, such as sanction checking, compliance issues, KYC matters, TBML, IFRS, and what have you, are actually post 9/11 rules and regulations. 

In this day and age of technical innovation and increasing cost cutting exercises, the tide is turning slowly but surely. It’s about time the ‘Lost Boys of Trade Finance Land’ grew up, and rather sooner than later, and guess what? It’s happening as we speak. We see we are no longer the only kid on the block. And this is good, it keeps us sharp.

Why is it that Trade Finance Operations got away without innovation for such a long time?

If you come to look at it, the real last revolutionary breakthrough in the world of Trade Finance has actually been the introduction of S.W.I.F.T. back in the 1980’s. Other than that it is pretty much the same game since 1865… But I can hear you say, “What about blockchain?” Fair enough, but let’s park that for a moment.

In general, on the one hand you have the traditional software developing companies of legacy systems, focusing on optimizing the operational status quo, not contributing to any basic changes of the way Letters of Credit, Guarantees and/or Documentary Collections and their paper-based presentations are being handled. On the other hand, there are the electronic document propagandists and the specialized sanction check offerers and publishers of lists trying to cover the increasing checks and regulatory demands' landscape.

Can you spot the gap? In our opinion, any revolutionary change in the market needed to combine all these elements in to one, dedicated, future proof vision and approach and it should be easy to use and implement without endless projects that only keeps the consultants companies happy. And then there is still the elephant in the room, the ancient craft of building and maintaining Letters of Credit (be it standard-, transactional- or back-to-back transactions), and the ever so infamous ‘Letter of Credit document checking’ which has never been really challenged by any system. (red.: the ICC UCP/ISBP based checking of documents)

There are initiatives that focus on electronic document creation and handling, which is an exciting turn in the right direction. But exciting and noble as it is [we are actually partners with the leading company in this field), without a doubt this will be a slow and painstaking process that will not be commonplace in the very near future.

Until all parties in the TF chain (The buyer, seller, manufacturer, trader, shipper, agent, transporter, railway company, airline, (customs) authorities, banks etc.) have agreed, accepted and implemented a Global standard for electronic document handling, much water will have passed under the bridge…..

This is exactly where I think Block chain and Smart Contracts will face the same stumbling blocks. So, until the market is ready to fully ‘go digital’, we are faced by the operational hazards of paper-based documents handling, in all layers of the Trade Finance chain for the next ten to fifteen years or so, in a market where demands in compliance, regulatory requirements, TBML monitoring, KYC and demands for business analytics will only increase. And that’s where we come in.

But how difficult could it be to optimize and automatize long known banking processes?

As much as we live in the 21st century, where our smart phones effortlessly perform actions once only possible in 'Star Trek', Trade Finance automation remains a medieval matter in a world that remains predominantly paper based.

Currently, most of those documents [if not all…] are being checked with the unaided eye to determine whether parties, goods and/or shipments comply with all requirements. Be it under Letter of Credit terms, or be it for compliance, KYC, fraudulent behavior and/or sanction purposes, each and every document is manually being looked at and judged upon its contents. Possibly even more than once by different people in different departments for multiple purposes. Needless to say, that this is an extremely time-consuming, complex and error prone endeavor, resulting in high numbers of operational costs, loss of business, or even result in governmental fines

We are all aware of the astronomical penalties that well renowned financial institutions were fined with, just for missing a dodgy party here and there in one of their transactions and were held accountable for this oversight. This simply cries for automation and traceable audit trail recorded checks.

This painfully reflects why so many efforts to innovate have been unsuccessful; They are initiated for and by banks mostly.

They are the ones suffering from the operational red tape, the extensive sanction- and compliance checks and what have you.

In my opinion the key factor to establish successful ‘change’ in the market would be a solution that serves both the financial institutions as well as the regulators and authorities, but most and for all the bank’s customers; The Producers, Exporters, Importers, Traders, Shipping Companies, Agents etc.

Only something that ‘fits-all' and does not create huge investments or too much adjustments to the current way of working, will have a change to succeed. Not only should this be a solution that will offer an answer to today’s issues, but at the same time is ready for the future. (Electronic Documents, Block-chain, Smart Contracts).

It was our strong believe that the first step providing an answer to these issues, was transforming the stream of paper documents into digitalized (carefully analyzed) data. It is the very philosophy with which we start building Trade AI

We fuse the best of both worlds and try to build a bridge to the future with a firm foundation in today’s processes. The business can keep their paper-based flow intact, and the banks can still digitalize the data and perform checks and analytics on it without having to mobilize dozens of operational people for manual time consuming and error prone checks.

Without getting too technical, what is the basic concept of the Application?

When we say: ‘paper to data' we are entering the wonderful world of OCR (Optical Character Recognition). A good OCR application is essential for any digital processing of paper-based documents. But this is just the beginning. By the way, to avoid any misunderstandings, we are not providing the OCR technology, we can basically integrate any existing machine out there. From ABBYY to IBM.

Now transforming paper into digital data is one thing (and this is exactly what OCR does) but what OCT does not do for you, is to provide what kind of data you are looking at, or show you if there are any parties/entities in that piece of text, or countries, or hazardous goods or anything else for that matter. OCR simply provides a string of raw data. So, once the OCR done its job, one still needs to ‘make chocolate’ of the actual raw data provided.

This challenge has been picked up over the last few years by a number of big players in the market of OCR and document processing, however they all took the same approach and faced the same major hurdle: Templates, and consequently presentation of standard documents to ‘intelligently' read the documents and store the data in to data types that makes sense. [Entity, Country, Amount etc.]

Now we all know, that if there is one thing that is not common in Trade Finance, it is standard documents and templates, Au contraire mon ami, I might say.

Each day different documents from different parties from all over the world turn up on bank’s desks and none of them in the correct template that was build.

No surprise though, that the toy factory in Bangkok, the large oil trader in London, the shipping agent in Alma Ata and the fortune cookie producer in New Jersey all use their own documents, made up of different templates, fonts and/or printers.

And yet all of those documents need to be checked and processed by the banks, as soon as possible, and without any time to create and adjust complicated templates that need to be build and maintained by already overloaded IT Departments.

How did you approach this challenge, and why did you succeed?

We acknowledged this real challenge quite early on in the process ‘How to read random documents which are totally unknown to the system’, to which the short answer is relatively simple; We decided we needed to build our own strategy recognizing the data from the OCR, and the ability to learn from it.

A four letter word…. MLAI (Machine Learning and Artificial Intelligence ) and I can proudly say that we actually found a custom innovative way to do exactly this. Detect parties, locations, goods, countries, SWIFT/BIC addresses, Vessel names, Ports and what have you, and then perform all kinds of checks on this obtained structured data by means of specific rules.

And without getting too technical and without giving away the sensitive information of our application, I will suffice by listing the main functionalities we are currently offering;

• Auto detection of the number and type of Document(s) uploaded and the number of pages per logical document [Doc.Classification]
• Data collection from the determined Logical Documents [Data Refinement]
• Screening against blacklists [public and/or custom lists] of the gathered data. [Reverse and Direct Search]
• Screening for TBML/AML, and Compliance Checklists of the gathered data.
• Create custom Questionnaires with Customer specific questions which can be automatically checked and answered by means of the collected data and custom black and/or reference lists.
• Common Consistency Checks between uploaded Logical Documents
• UCP/ISBP Letter of Credit Document checking based on the L/C|SWIFT [Contract] and the documents presented.
• Vessel Check . Intelligent checks on the Vessel, its origin, owners and routes. All done automatically based on the Vessel detected in the presented documents.

And last but not least, we are proud to be able to state, that all our checks can be performed against 100 % of the gathered data, the exact problem all manual-based processes suffer from.

It’s not rocket science to realize that what can be achieved with paper documents being fully digitalized, categorized, read and interpreted by intelligent software, this principal can be applied to many other areas in the Financial World.

As we speak, we are developing modules for: KYC, Insurance business, Loans, Mortgages, Finance Products and we are working on a 'Light' on line version that offers Corporate Customers the possibility to do easy and quick checks on new Customers, or pre-check their L/C documents.

How do you see the future of Trade AI?

Extremely bright. But if there's anything we can learn from the past, it is that it is hardly impossible to predict, even, the near future.

We are currently looking to expand our business in to other areas such as Health Care sanction checks, Public Transport, Municipalities and the like. Not only do we follow the developments in the market, we also try to stay ahead of the game or even lead the way and dictate the game. We also encourage our customers to confront us with new challenges, no matter how complex. We like to call ourselves and our products ‘visionary and progressive’.

So as far as we are concerned, the Sky is not the Limit, but only the beginning…