Eight opportunities for automation

• The automation of routine work has accelerated in recent years
• The growth of automation could lead to the development of a variety of services to assist in-house trademark counsel
• Computers will assess the chances of success and levels of risk, using ever more sophisticated data analysis

In the trademark field, the automation of routine work has accelerated in recent years, as trademark office records have become widely accessible, enabling free services such as TMview, as well as more accurate and detailed commercial searching and watching services. Processes that were once labour-intensive, such as changing names on registrations or renewals, are now done almost automatically. The development of these services is discussed in more detail in our report on trademark service providers.

To be successful, in-house counsel must embrace automation and the benefits it brings: efficiency, speed and accessibility. For some, this will mean a fundamental change of role, as paralegals or assistants are no longer required to perform tasks that can be done by a machine. Even for more senior staff, automation may bring new challenges. Alan Minto provides one example – a career trademark attorney who has worked at companies including Unilever, Dr Martens and Gallaher, he spent 10 years as IP vice president at JTI, in charge of the company’s global trademark portfolio. In September 2018 he moved to a newly created role as vice president for innovation and productivity, focusing on developing and acquiring software tools throughout the legal and regulatory affairs department.

The growth of automation, including AI tools such as machine learning, deep learning and neural networks, could lead to the development of a variety of services to assist in-house trademark counsel. With much of this technology still undergoing beta testing, and given that some projects are likely to be confidential, it is hard to predict exactly where we will be in 10 years’ time. But the research behind this report suggests that the following ideas illustrate the kinds of tool that could be developed:

• Central repositories for internal legal records – these could include opinions from outside counsel, interactions with IP offices, relevant court judgments and examples of trademark use. Advanced search engines would be able to find and identify relevant records instantly. One big advantage of such services is that they would prevent the duplication of work when staff move on or are absent; everything is stored on a server (or the Cloud) rather than in a filing system or in someone’s head.
• Similarly, improved searchable databases of external counsel and other service providers –existing record systems could be enhanced to track all interactions with outside advisers (eg, enabling the analysis of bills, level of responsiveness and success rates). These could also incorporate some form of scoring outside advisers, which could feed into decisions about retention and appointment. One characteristic of large trademark portfolios is that they require a lot of external counsel to manage, so trademark work is perfect for this kind of service.
• Trademark searching – AI has the most obvious potential for availability searches. Some tools are already doing this, but they can be improved, particularly for coverage of non-traditional marks and in certain jurisdictions. Some practitioners are sceptical about whether computers will ever be able to make the subtle assessments needed about the similarity of marks and goods or services, but developments in fuzzy logic could prove them wrong. Even if a computer cannot give a 100% correct answer, if it can get 90% of the way there, then that would save a lot of time and work.
• Drafting applications – one of the fastest growing areas in AI is natural language processing. This includes machine translation, which is already providing benefits to trademark practitioners. Combined with other AI tools, natural language processing could simplify and speed up the steps from brand creation to trademark registration, including monitoring and responding to interactions with IP offices.
• Cancellations and oppositions – although requiring more legal analysis than prosecution work, administrative procedures such as cancellation actions and opposition are also ripe for automation. IP offices point out that filings frequently include errors (eg, regarding dates, descriptions of goods and services and reference numbers), which computers could eliminate. “It’s realistic to have the first response to an opposition drafted by a machine. It can be done from a template and sent for review,” says one in-house counsel. Bear in mind that offices in all EU member states must provide administrative trademark proceedings by 2022.
• Evidence – AI tools have great potential to be used in trademark litigation, particularly for sourcing, compiling and organising evidence; for example, tools could easily scan the Internet and social media for examples of consumer confusion and also make it much easier to find precedents and assess their relevance. However, courts are likely to be wary. To take a simple example, the results of a Google search may *prima facie* indicate a likelihood of confusion when certain search terms are entered, but a judge may ask: to what extent are the results skewed by the individual’s browser history? Do they represent the experience of the average consumer? Over time, specialist tools will need to be developed to address such questions.
• Agreements – this is one area where trademark law may benefit from advances in other fields. There are many tools being developed in other legal disciplines such as corporate, banking and property law to develop so-called ‘self-service agreements’, saving hours of work. This investment ought to be easily transferable to trademark licences and coexistence agreements, which can be replicated with minor changes for each case.
• Making decisions – the big question that keeps professionals awake at night is whether a computer could ever be trusted to make a decision unaided; for example, whether to continue or abandon a proceeding, sue a competitor or decide whether a mark is distinctive or whether two signs are confusingly similar. This would pose a direct threat to the existence of trademark practitioners (and examiners and judges, for that matter). However, it seems unlikely that computers will be capable of making such decisions in the near future – for legal and business risk, as much as technological, reasons. It is more likely that computers will be used more frequently to assess chances of success and levels of risk, using ever more sophisticated data analysis.

What all these tasks have in common is that they include some element of routine or predictable work. “75% of what in-house counsel do is fairly routine work,” says one former head of trademarks. “That doesn’t mean it’s unimportant, but it is susceptible to being done better by technology.”