Expert

Prof. Jan Scholtes about the AI revolution in M&A

Artificial intelligence (AI) is ever more frequently being applied in mergers & acquisitions. Companies, law firms and private equity investors use robots because they work cheaper, more accurately and faster than humans. As Jan Scholtes (ZyLAB) and Nancy Brewster (Legadex) have discovered, buyers are increasingly imposing searchability of a data room as a hard requirement. “The use of AI lets you retrieve a huge amount of information in a short period of time.”

Text Michiel Rohlof Photos Geert Snoeijer

An interview with Jan Scholtes (Zylab) and Nancy Brewster (Legadex)

Trainee lawyers fearing to get bogged down with a lot of due diligence research in their first years at a firm can breathe a sigh of relief: the use of artificial intelligence (AI) is now more or less commonplace and real manual work is becoming a thing of the past. Techniques used in data science and AI are helpful tools in situations where there is a lot of unstructured information or where information has to be collected from different sources and then correlated. What is more: in deals involving venture capital or private equity investors, searchability of the dataroom increasingly often qualifies as a hard requirement.



Jan Scholtes is professor Text Mining at the AI-group at Maastricht University and Chief Strategy Officer at eDiscovery expert ZyLab. “Investors are saying: I want the entire data room in PDF-format or the deal is off. That’s when we use AI for fact-finding: retrieving the correct information, filtering out everything that’s irrelevant and missing as few relevant documents as possible. What investors do is look for inefficiencies and improve a company’s operations from within. If you want to be able to do so quickly and efficiently, AI is the way to go.”

Appealing proposition
AI is not only an appealing proposition because it offers speed, efficiency and cost savings, but computers are also better when it comes to accuracy. “Computers too may supply incorrect information at times, but they are consistent when they do that; this makes errors easier to trace and correct. We humans are inconsistent in our inconsistency, which is much more complicated. LawGeex recently conducted a study in which they instructed top-class lawyers and computers to check contracts. This resulted in an accuracy rate of 94% for the computer and an average accuracy rate of 85% for the lawyers – the best lawyers were 94% accurate and the worst scored an accuracy rating of 67%. The computer got the job done in 26 seconds, while it took the lawyers an average of 92 minutes to finish. Those are the hard and fast facts.”

AI offers speed, efficiency, cost savings and accuracy: on average, lawyers score 85% on accuracy versus 94% for the computer

Nancy Brewster is teamleader Transactions at Legadex. She describes a recent example of how AI was successfully used in an M&A transaction. Commissioned by a seller, artificial intelligence was used to make a retail chain’s real estate portfolio more transparent and to increase its searchability to allow its sale. Brewster: “To that end, data covering a long period had to be collated from different sources. The seller was adamant that the portfolio should be presented as a complete and coordinated set of activities. We made this happen by letting the software search all available data and subsequently making it draw conclusions in a report. That may sound obvious, but it’s revolutionary in its execution: you use your data sets to train the algorithms such that they recognise patterns and become smarter over time. This makes the process a lot easier the next time. Without the application of AI, this would be impossible or unaffordable.”

Highly organised list
Another recent example is the sale of a loan portfolio by a financial institution. Brewster: “When it comes to loans, a file must be complete. We agreed in advance when we would consider a file to be complete. We then went on to conduct in-depth research using AI and the computer collated relevant information at file level, using email traffic and different document formats at different storage locations. This resulted in a highly organised list of incomplete files and the workfloor took action immediately.”

Scholtes states that modern AI is based on machine learning, in which process computers learn to recognise patterns in data sets and print out relevant reports. This is easier to apply in practice than traditional AI, which is more rule-based. To illustrate: Watson, an example of traditional AI, is difficult to maintain without technical knowledge. Modern AI, which is used by Netflix and Spotify – services that many people use on a daily basis – for instance, allows computers to work with relatively small data sets. As a result, customers have easy access to AI. Scholtes: “Using modern AI, you simply share a lot of information with the computer, which then recognises patterns automatically based on the words in the documents. This makes things much easier for both buyers and sellers in due diligence reviews. Potential problems are a lot easier to trace, possible deal breakers are recognised and no more skeletons will come falling out of the closet.”

Personal data
Technology also helps companies be compliant with the new European privacy law: the General Data Protection Regulation (GDPR). “Large quantities of data, including contracts and email traffic, can be scanned automatically for personal data,” says Scholtes. “This allows you to see right away what personal data is stored where. That information can then be anonymised or pseudonymised, if required.”

Computers can do a great job organising, filtering and arranging, but we need people to draw conclusions

In short, AI will take over many standardised tasks, but human verification will always be required. Brewster: “Computers can do a great job organising, filtering and arranging, but we need people to draw conclusions. In a due diligence review, for example, you check which advisers require which information. Then you look at what the core documents in a file are. All draft versions of a document may contain incorrect information; with this in mind, you should only use signed copies for your research purposes. By properly defining beforehand what it is you want to know exactly, you remain in control.”

Major advantage
Scholtes expects AI to evolve significantly over the next few years. “We’re already seeing that companies and investors continue to use the technology even after a merger or acquisition: searchability of company data is considered to be a major advantage and companies effectively have a data room ready-to-go at any time. In the context of the GDPR, insight into data is also hugely important. Some private equity firms and companies are even using AI to measure the productivity of their advisers. The possibilities are endless.”