Afcacia had the opportunity to interview Alioune Ciss, chief executive of Webb Fontaine Group (pictured)on how Artificial Intelligence is boosting international trade. Alioune is key to the strategic drive, focus and direction of the Webb Fontaine brand. He has beeb managing world leading trade reform digitalisation projects across the globe with key governments and partners, through innovative technology. Alioune is thought leader within the Trade and Customs field, strategically bringing together new technology and ideas to practically implement and change mindsets. He has managed and delivered some of Webb Fontaine’s most ground breaking projects such as smart fraud detection for the Nigeria Customs Service. He has partaken in many panel discussions with the World Customs Organisation, United Nations, AFRITAC (IMF Regional Center for West Africa), WB/IFC and African Shippers Council Organisation. Prior to his work at Webb Fontaine, Alioune spent 24 years at the United Nations Conference on Trade and Development where he was for the last 10 years the Project Coordinator for Africa and the Middle East. Here, he caught up with Faustine Ngila who sought to understand how the company deploys AI to ease painpoints in international trade.

  1. What inspired the founding of Webb Fontaine?

Webb Fontaine was built on the belief and focus of looking beyond barriers to shape tomorrow’s international Trade. We imagine a world where Trade is carried out seamlessly across borders, a world where people and businesses can exchange goods and know-how in the most hassle-free way possible for the benefit of their community.

  1. What does it do? In which African countries is it present?

Webb Fontaine is a world-leading technology company re-shaping the future of Trade. We provide industry wide solutions to accelerate Trade reforms and modernization.

The company uses state-of-the-art technologies including Artificial Intelligence to enable countries to emerge as leaders in the future of Trade.

We have a large presence across Africa including: (1) Nigeria (2) Benin (3) Cote d’Ivoire (4) Ethiopia (5) Central Africa Republic (6) Congo (7) Guinea (8) Niger (9) Egypt

  1. What subsets of AI does Webb Fontaine use to aggregate Trade? How does that work?

As the international Trade process is a vast and complex network consisting of varied activities and sub-processes and thanks to the fact that Webb Fontaine is active in the whole spectrum of it, we are using a large set of Artificial Intelligence and Machine Learning models.

With ML we aim to simplify and automate bottlenecks and repetitive activities affecting stakeholders throughout the industry.

For example, we are working with international Trade documents such as certificates of origin, bills of lading and invoices to solve issues such as forgery and manual data entry. Our models use image processing, optical computer vision (OCR), named entity recognition (NER) and classifier models to read, understand, detect fraud and extract trade-applicable data. Our AI document assistant first makes traders more efficient by automating their declaration data entry activities.

Then, after the document is submitted to Customs, our fake document detection algorithm is able to flag images that have been tempered with or have an unusual format for a particular exporter. The combination of these two approaches allows the industry transform paper documents to electronic data in a fast and secure way.

Another example, and a widely known one by both traders and customs agencies, is the complexity of the international HS code nomenclature and knowing which HS code to use when declaring, evaluating and classifying goods.

Using natural language processing (NLP) and text classifiers, we built a google-like search engine for HS codes that returns the appropriate code for any good based on its description, the model is also used for automatically classifying goods in certain contexts.

The model also understands and uses pricing data in order to predict prices of products as to automate the valuation process. Our solution improves the communication between companies, traders and customs in that serial numbers, technical codes, brand names and vague descriptions are clearly translated into HS codes.

As a result, the overall clearance times of our clients are decreasing as traders are using the right codes faster and, if they are not, customs inspectors are rapidly finding that this is not the case and are able to correct any inconsistencies.

In regards to the problem of fraud and tax evasion, our Smart Fraud Detection solution help our clients detect and analyse fraud. With high accuracy, our models detect fraud trends, outliers and asses the risk profiles of multiple trading entities.

Our solution holistically assesses declarations and assists inspectors by taking a decision whether to invest time in inspecting a declaration. To make our models transparent for decision makers and ensure that inspectors are confident with their assessment, we display all the factors constituting a decision and their corresponding weights.

Our Smart Fraud Detection solution is augmented with the results of our other models such as fake document detection, HS Code verification and our Price Prediction to evaluate the accuracy of declared data. Clients using Smart Fraud Detection are able to optimise the time taken and the number of inspections to increase fraud capture efficiency.

  1. How does Webb Fontaine protect the user data it collects across its solutions? How does it address customer data privacy?

Given the sensitive nature of government Trade data, privacy and data security are a key focus for Webb Fontaine. Contrary to data aggregating companies whose aim is to amass as much data as possible for reselling purposes, our goal is to ensure the optimal efficiency of every single client and solve problems in their local environments.

Our commitment to data privacy and security starts with the data centres of our clients, where we ensure that private networks, access control and secure network policies are implemented. We anonymise data before processing on our servers. Additionally we do not use cloud services which means the anonymised client data remains with our AI research teams locally.

The AI solution models are trained with a single customer data set at a time which prevents data from being shared between clients by the models. When we need to reuse models already trained for a specific purpose, we use transfer learning which is a machine learning technique enabling the reuse of a trained model without using or accessing any the original data.

  1. What role does AI play in international Trade?

Today, AI is in its nascent stage in international Trade. We at Webb Fontaine believe that AI will revolutionise Trade and play a key role in the future of the sector. With millions of transactions on a daily basis based on standardised workflows and document types, AI will transform the way governments Trade with each other.

In the domain of Trade, AI has the potential to be a game changer with its demonstrated ability to automate repetitive, manual tasks such as visual inspections, data entry, data verification, data classification and fraud detection. The financial sector is similar in terms of volumes and types of transactions and AI is thriving there.

Both traders and customs administrations are spending significant amounts of their time and resources in performing manual operations today, tomorrow, AI will be carrying this load for them so that they can focus on improving their operations and optimising clearance times.

Risk management is another domain where governments invest heavily and carry out repetitive inspections. Artificial intelligence models have strong fraud detection capabilities and will serve as a trusted assistant to inspectors and administrators.

This is why we are investing in and pioneering AI solutions. We have a dedicated team of AI experts, often holding PhDs in mathematics, working exclusively on our AI capabilities in the context of Trade. Today, we have developed and deployed numerous solutions and already see strong gains in productivity and revenues in the customs administrations working with our products.

  1. What challenges does Webb Fontaine face in accessing African markets?

The overall workplace of digitalisation as well as local internet connectivity are two specific challenges related to African markets.

With that said, starting fresh, not having legacy systems and ageing operations is actually an opportunity for many of the market players to implement modern, best-in-class systems. We believe that the digital capacity of African markets is only going to expand over time and we are excited about it.

Artificial Intelligence models are fully data reliant and another challenge is data quality and content. With that said we have a team of data quality experts and data annotators who prepare data for usage within AI models.

Artificial intelligence transfer learning for specific regions in the African markets is also available. Administrations without strong data capabilities can also benefit from existing models developed in their region.

We have successfully implemented production models working on hundreds of thousands of and we are confident that we can drive the implementation and usage of AI in the context of Trade throughout Africa. Thanks to our local understanding of the African market and are technical capabilities, we can deploy AI solutions within a short time.

  1. What are your general comments on ethical AI in Trade? What dangers are imminent if AI remains unregulated?

AI models used in international Trade solve clearly defined problems and are trained on expertly narrated data. At Webb Fontane, we believe that until the AI field is mature enough on a global scale, AI should be used as an assistant, or one more opinion, while having government officials in control of decision making. This is the approach we are taking with our AI products and their implementations.

Having already closely collaborated with several administrations in the implementation of artificial intelligence, we are ready and willing to drive the implementation of AI within the region by providing advice and assistance in finding the subtle balance between technological progress and regulations and ethics.

Outside of the context of trade, we all have seen the imminent dangers of AI solutions based suboptimal and biased data. Clearly defined regulation with regards to data sources, data analysis, model development and model application is of great importance.