Predictions for LegalTech in 20193 min read

This year has been saturated with events in LegalTech: from the first ever Global Legal Hackathon to the launch of PrivacyPolicyCheck.Ai, first ever client-oriented legal AI developed by a Nordic law firm in-house. On the trends side, we have been witnessing the ever-growing use of AI in day-to-day work at law firms, which demonstrates the decrease in technological skepticism and uncertainty among lawyers and increase in common understanding that AI enables more efficient workflows. Having seen more practical examples of AI applications in law, legal professionals now tend to be curious about LegalTech, which is an improvement comparing to 2017 which demonstrated mostly awareness rather than interest.

There are still two months till 2018 ends, but, considering the usual business activity spike in early autumn which normally determines how things will be working out till the year’s end, it is a good opportunity to try to predict what to expect from LegalTech next year, in 2019.

AI. The hype surrounding AI will keep building up, fueled by a better access to “tangible” applications. Expect more cooperation on large projects between laser-focused LegalTech companies, each doing its part but doing it well. For the first time, prototypes of legal AI “rewriting” legal documents will be developed, built upon the well-tested document review infrastructure. These prototypes will prove themselves useful fast, although they will still be inaccurate in 10-15% of the time and limited to a specific document type. New kinds of neural networks and data extraction techniques will be developed, which means the increase in patenting of LegalTech applications.

Blockchain. The blockchain hype will last till mid-2019, followed by its gradual reduction due to the growing understanding that using blockchain is not practically justified in 99% of cases. The remaining 1% of apps will be very successful in penetrating the market, partly because of the above-mentioned hype reduction and therefore less “noise” to break through.

Automation. “Automation is not AI” discussions will be more frequent not only because of the widespread proliferation of legal AI solutions but also due to the shrinking gap between what used to be programmed explicitly (automation), such as key data extraction using hard-coded patterns, and what can be “figured out” on its own (machine learning). With new techniques and types of neural networks developed, the need for hard-coded software will decrease over time, with automation being gradually replaced by self-sufficient AI tools.

Funding. A more robust proof of the practical utility of AI solutions combined with the ever-increasing hype will drive more investments to LegalTech. This year has shown some record numbers, such as the recent $50M investment into Kira Systems, the largest ever for a Legal AI company. Expect these records be beaten in 2019, and not only by private actors.

Open-source. Despite a Bitcoin-like evangelism in the open-source domain, the LegalTech backend will keep doing its magic over account-protected web-interfaces and APIs. This is mostly due to the need to deliver higher ROI while it is still possible in the legal industry which is undergoing rapid transformation. While a few LegalTech libraries are emerging already now as an open-source product, from my experience such solutions are very limited in their functionality, not flexible (as every open-source should be), and are build to serve a marketing purpose.

Market overall. The pace of disruption will accelerate. Traditional legal markets will be heavily influenced by successes in neighboring jurisdictions which are more innovative and adaptive to change. Unwieldy LegalTech landscapes will become more vibrant mostly with the help of outside actors, who already have a good track record in practical innovation in their home markets.

How close the above predictions are to reality – time will tell. For now, what we can know for a fact is that a neural network’s main function for making a decision is usually named “predict”.

Sergii Shcherbak