How AI helps cut your costs – but not your confidence – during due diligence

Mergers and acquisitions are both exciting and exhausting. Tight deadlines, limited information and large amounts of money at stake mean that there is significant pressure to gain confidence in a deal. This is especially the case when intellectual property is the primary asset changing hands. However, AI and machine-learning software could be set to transform this space.

Assessing the risk of infringement, or acquiring invalid IP rights, requires much more than due diligence boxes. Whether an IP attorney is in-house or outside counsel, they are responsible for communicating the risk to business stakeholders. Speed, accuracy and clarity are all crucial, and one of these often comes at the expense of the others.

Transactions take place within a tight time frame; therefore, understanding the IP implications of the deal is critical. While you need to get up to speed with the company’s intellectual property fast, it is also vital to understand that of their competitors and how combing your product offerings could affect your technology landscape.

Patents do not have a uniform taxonomy or nomenclature of technical terms. It is not as simple as asking the USPTO whether your product infringes on that of your competitors. Instead, it is necessary to search the patent database of every country where you currently (or may someday) do business, for any patent that might be relevant to the company that you are acquiring.

For a small team on a budget or deadline, keyword searches are not sufficient to accomplish this. Even with technical expertise in the subject, you may not know of every synonym to search. Using machine learning (ML) or AI is essential to identify all potentially relevant patents and assess the risk they pose. Combining human expertise with ML or AI is an effective recipe for balancing thoroughness, speed and budget. A variety of patent research databases and software tools are available for this type of analysis, and they are becoming more accessible for lawyers every day.

ML, as an example, enables conceptual searches on customised search indexes. One type of conceptual search index, Latent Semantic Indexing (LSI), converts chunks of text (eg, patents) into mathematical vectors and places those in a multidimensional concept space. Subsequently, the machine generates a ranking for how close or far a given patent is to a known patent, without relying on keyword or Boolean search alone. Combining human judgement and a strategy such as LSI with other types of ML or AI drastically reduces the time required for a search, while simultaneously lessening the risk of missing critical references.

Traditional Boolean searching may yield thousands or tens of thousands of hits – far too many to read in detail. It is thus critical to combine multiple strategies in order to achieve a balance between thoroughness and speed.

The completion of an acquisition teaches a team that technology is a powerful force multiplier. While ML and AI are not silver bullets, they form essential components in of a practitioner’s toolset. When combined with thoughtful keyword, Boolean and data visualisation strategies, they enable teams to make big decisions during diligence while reducing risk, staying within budget and finishing on schedule.