default logo

Automating Legal Advice: AI and Expert Systems

In 2015, we regularly read about automating legal advice with artificial intelligence (AI), especially with IBM Watson. In my view, the AI smoke – at least as described in many reports – exceeds the fire. In fact, I cannot name a single large law firm that has deployed a Watson system.

Yet, as I explain below, expert systems, a branch of AI, are actually being deployed. Several firms have licensed expert systems to automate advice or intake. I start with some background on AI before turning to recent expert system developments.

Artificial Intelligence (AI) in the Legal Market – a Recap

I first became aware of legal Watson at the 2014 ILTA conference. The private track of large firm CIOs heard a presentation on IBM Watson. Based on second hand reports of it, I wrote Meet Your New Lawyer, IBM Watson. It explains my skepticism about Watson’s immediate prospects in the legal market.

Intrigued nonetheless by the potential of IBM Watson, I attended World of Watson, a slickly produced event at the uber-hip Brooklyn Navy Yard in May 2015. (See my live IBM Watson posts.) It was fun, especially the meet-up of Big Law lawyers and managers that I organized. But at the end of day, I gained little insight into the potential for AI in legal.

Later that May, I wrote Legal IBM Watson: Business Model and Reasoning Modes, which explains the business model challenges of deploying Watson: a big time investment and much content. After listing the three uses of Watson I could find in legal, I closed noting

Other ways exist to automate answering legal questions. How do we assess the cost/benefit of Watson with alternative approaches? For example, expert systems from Neota Logic answer questions via an interactive Q&A session.

I will return that alternative below. Forward to 2015 year-end when economists Dana Remus and Frank S. Levy published Can Robots Be Lawyers? Computers, Lawyers, and the Practice of Law (SSRN,30 Dec 2015) by. I found it eye-opening. So did New York Times reporter John Markoff, who wrote The End of Lawyers? Not So Fast (4 January 2016).

Remus and Levy conclude that the impact on legal employment of AI will be quite limited for the foreseeable future. I found their paper a good antidote to hyperbolic press in 2015 suggesting that AI would sweep the legal market.

All this said, I do see potential for AI in the legal market. And I await eagerly the release of Kim by NewLaw firm Riverview Law (7 Dec 2015 press release). It is an AI system not based on Watson.

Also, lost in the 2015 legal AI hype is that expert systems are a branch of AI. Michael Mills, co-founder of Neota Logic, explains this in detail in Artificial Intelligence in the Law – The State of the Play in 2015. His published diagrams shows the many branches of AI:


 Artificial Intelligence (AI)

Legal Expert Systems are Here Now

I have a long-standing interest in legal expert systems. For over two years, ending just as this century began, I worked for Jnana Technologies, the developer of an expert system platform targeted primarily at the legal market. I built or helped build legal expert systems that were deployed at that time. Subsequently, Neota Logic acquired the underlying technology of Jnana. (I have no financial interest in Neota.)

Now I return to the point about alternatives to other types of AI and alternatives to Watson. In my view, expert systems are here and now. That does not mean easy. Building them requires deep legal expertise to start and then for on-going maintenance. Moreover, the business model challenges law firms:Why would lawyers forgo billable hours to build a system?

Because of the billable hour conundrum, at one time, I thought big legal publishers might build expert systems. If they do, I am not aware.

Some law firms seem to be overcoming the billable hour block. Firms today face a litany of challenges: low realization, flat demand, market share battles, and clients pulling work in-house. In response, some forward thinking firms are moving beyond the billable hour, at least for purposes of offering automated advice.

In December 2015, I spoke with partners at two large law firms that are developing systems based on Neota. Martin T. Tully of Akerman co-spearheads his firm’s development of the Data Law Center. It is, according to the website, a

“first-of-its-kind legal product developed in collaboration with Thomson Reuters to support the compliance strategies of large companies. The Akerman Data Law Center will provide corporate counsel and compliance officers with a timely trove of state and federal data laws, regulations, and legal insights. This product will deliver tailored research, multi-jurisdictional surveys and regulatory gap analyses in a wide array of data and privacy risk areas empowering clients to quickly and cost-effectively understand and handle routine compliance matters.”

Martin shared that its development stemmed from the firm’s R&D Council. The Neota Logic system will first offer (it is not yet released) services for data security and privacy laws; eventually it will cover other data issues, for example, records management. Both Akerman and the Pangea3 managed services division of TRI will keep the resources and system logic fresh. Martin expects the system will eventually answer 80% of subscriber questions; the remaining 20% may require traditional legal services, which are planned to be bundled in the offering.

David W. Simon of Foley leads his firm’s development of the Foley Global Risk Solution, based in part on Neota Logic. Mr. Simon explained that this system meets

“the FCPA/anti-bribery compliance needs of companies with international business operations that may not have the internal resources and personnel to develop and implement an effective compliance program on their own.  FOLEY Global Risk SolutionsSM (GRS) is an award-winning, fully integrated FCPA compliance solution that addresses each of the “hallmarks” of an effective anti-corruption compliance program identified by the U.S. Department of Justice and the Securities and Exchange Commission”

The system is still in launch phase but has six clients signed up. David reports though that he has had over 50 meetings with prospects to demo the system – a far higher level of interest than he has seen for conventional practice pitches. Over time, it becomes more powerful because the Q&A database that drives some answers grows. The target market is mid-size companies with without in-house counsel. For the moment, the firm will make it available only to clients but will eventually consider a subscription model for non-clients.

Foley and Akerman are not the only firms that have licensed Neota Logic and are building expert systems with it. Here are other instance with public mentions of uses of Neota Logic:


Neota Logic is not the only option for expert systems. I have not searched for the universe of available systems but am also familiar with VisiRule by Logic Programming Associates (link to demos of legal expert systems), Berkeley Publisher by Berkeley Bridge, and Oracle Policy Automation (used, for example, in the IRS Interactive Tax Assistant (ITA).)


When it comes to automating the delivery of legal advice, AI as epitomized by Watson has gained the biggest share of mind. I suspect – and you can evaluate the evidence above – that expert systems actually have a bigger share of market in legal.


Note: This article was first published by Bloomberg Big Law Business as Automating Legal Advice: AI and Expert Systems on 22 January 2016

  1. John McConchie

    Thanks for an excellent, up-to-date article on the current state of the art in expert systems. I think that the law firms that are pioneering expert system applications in the 21st Century (sorry to those who spent precious money on them in the 1980’s or so) have two upsides, one possible, one certain.

    The upside which is “possible” but not certain is that they will make money. Good for them if that happens. We understand in this society that pioneers often get slaughtered and their exploits benefit only the future generations who tend the graves while slurping ice-cream cones (although with this group of pioneers, there may be a bit of blood and bile, but no graves).

    The upside which, to my mind, is “certain” is that these firms will increase their knowledge and smarts in their chosen areas of expertise in leaps and bounds, and I bet it will come as something of a surprise to them because of how smart they already are. I am looking forward to reading some articles (could you write them please?) on how the “subject matter experts” involved in these projects would describe the experience of working on an expert system, however “simple” it may appear from the outside. Did they spend their time sitting at the edge of the pool spinning out “rules of thumb” for eager, ascetic knowledge engineers? Or did they have to burrow down into the bowels (sorry!) of their own personal knowledge and experience bases and, having found what they needed, methodically build a tower of logic and knowledge fit for a useful expert system? With sweat pouring down? If you get a chance, ask them: in retrospect, “do you think you were an “expert” before you did this? How do you feel about the extent of your expertise now?” The chances will be that the expert will agree that they never before had the degree of expertise they were required to have to build the program.

    In short, I think that the law firms that are building expert systems may not build an immediately prosperous business (although they may) but they will most assuredly improve their expertise in the areas in which they practice. Of anything, I am very sure of that.

    What gives me the gumption to express these opinions? I have no real sense of that. As a lawyer in a small firm who has found expert system development with an expert system shell to be a “fun” hobby (I kid you not!), I am a follower of the trends. I am working on stuff but it will happen or not in due course. I find articles like you have drafted to be fascinating. Thanks.