kompany uses AI to revolutionize ultimate beneficiary ownership discovery
In the world of due diligence, achieving full insight into ultimate beneficiary ownership (UBO) structures must rank as the most difficult challenge of all. Shareholder structures can be opaque, comprising nebulous networks across many jurisdictions.
To make it even harder, in some countries the state of the official registers themselves leaves a lot to be desired. Shareholder information tends to consist of unstructured data, with no consistency of how it’s collected, recorded or stored. I hear horror stories of entire compliance teams wading through assorted PDF formats, inconsistently designed forms, even low-quality scans of hand-written documents. Also, within the same country, different regional bodies can log data in varying ways.
Factor in too that the latest anti-money laundering legislation is insisting on more rigid UBO checks. While the EU’s 4th Anti-Money Laundering Directive (AMLD4) made banks’ and FIs’ checking of beneficial owner registers optional, AMLD5 makes it obligatory.
kompany operates a proprietary global live network to access company filings for over 110 million entities in more than 200 jurisdictions. The data integrity is guaranteed, time-stamped and direct from official primary sources. This makes the data legally reliable. And now, thanks to our innovative use of artificial intelligence (AI), we are applying the same rigor to unstructured shareholder documents, delivering audit-proof easy-to-use information. Our customers can verify not only the companies themselves, but also the people behind them effortlessly.
So how did we do this? By tackling the jurisdiction with the most challenging shareholder data first: Germany.
Germany, perhaps surprisingly for Europe’s biggest economy, does not have a centralized governance system for its company filings. Instead over 100 regional authorities manage their own local registers, each with its own form and preferred way of maintaining data. Considering the country’s commercial backbone, the Mittelstand, comprises around 3.5 million companies, this is a logistic nightmare for compliance departments, and UBO checks can literally take weeks.
Normally we would have dealt with this challenge via the use of deep learning and robotics, but because of the huge complexity of this unstructured data, we were forced to apply a neural network approach.
Working with the computing hardware of our strategic technology partner Oracle, kompany started with over 70,000 different models that were tested to read and interpret the German shareholder data. These were then whittled down to 640 which were subsequently put into AI training. By narrowing the models down even further, after two years’ hard work, kompany now has the tech logic to transform unstructured data into a usable structured format with 95% accuracy.
Based on a customer’s request, a PDF containing, for example, unstructured data as an image layer, is extracted in real-time from the respective primary source register. After applying optical character recognition (OCR) software and turned into text, it is then run through our in-house tagging and analysis engine. This is then automatically turned into a document in a consistently ordered format.
kompany is now excited to launch this pioneering functionality via its new UBO discovery tool. Visually displaying connections between companies, officers and shareholders, it helps reduce UBO investigation time by days, and is fully complaint with AMLD5.
An estimated 95% of all company information in registers around the world is unstructured data. We at kompany are helping our customers to make sense of it and create a transformative difference to the way compliance teams have been forced to run their operations up to today.
For more information or a demonstration of our new UBO discovery tool, click here.