A Legal Tech Tool That Is Actually Very Cool – But Does It Break Ethics Rules?

To date, much of what’s been characterized as legal disruption hasn’t impressed me as much more than a souped-up, techno-powered version of what’s been tried before and failed. Still, I’m no curmudgeon and every so often, I find myself excited about the possibility of tiny law applications like Shake Legal – an app to create quickie contracts for small matters where a lawyer doesn’t make economic sense) – and now, Legal Sifter.

Basically, LegalSifter employs technology to eyeball a contract, much in the same way that an experienced lawyer might if asked by a fellow attorney or a friend to have a quick look.  As described at Tech Crunch, Legal Sifter

use[s] natural language processing…and scans your documents (in Word format only, for now) and assigns them a score based on how favorable the terms are for the user. It also provides users with an explanation of the clauses and provisions in the document and suggests potential changes to provisions that are probably not in the user’s best interest.

LegalSifter was developed to review consulting contracts for designers, so the scores assigned are most relevant to these types of engagements. Nevertheless, as an aficionado of another type of service agreement – specifically, the Legal Retainer Agreement, I couldn’t resist running a sample engagement agreement through LegalSifter.

I used the Standard Hourly Billing Agreement below from Lawyers’ Mutual online Attorney ToolKit.

I converted the document to a word file and uploaded it to LegalSifter.  In under a minute, the site generated the following result:

Screen Shot 2014-09-02 at 7.13.06 AM800

Legal Sifter scored this representation agreement as an 87. According to the notes, the score is intended to measure “how fair the contract is” and explains that any score over 50 shows that the agreement leans in favor of the designer (or more generally, it would seem, the service provider or the attorney).

Legal Sifter doesn’t stop there, but offers suggestion ways to make the contract even more favorable.

Screen Shot 2014-09-02 at 7.13.29 AM800

At least two – the limitation of liability and attorneys fees clauses are potentially inapplicable: lawyers generally may not limit malpractice liability ( e.g., Bar of City of New York Formal Opinion 1995-7) and traditionally, do not include fee-shifting provisions in representation agreements since if the firm would not be entitled to fees if it prosecutes the case itself. And while lawyers do not typically “disclaim warranties,” LegalSifter’s recommendation is a good reminder of many jurisdictions’ ethics requirement that lawyers expressly state that past favorable results do not guarantee the same in future cases.

But can LegalSifter be considered “the practice of law?”  More so than other legal innovations, as I see it. LegalSifter renders the same type of feedback as would an experienced lawyer eyeballing a contract – and indeed goes a step further, advising users on ways to further improve their agreements. In this regard, LegalSifter has far more capability than third-party form fillers like LegalZoom, which a South Carolina court characterized as operating like an automated mail-merge program. Whereas many of the legal innovations we’ve seen to date at best, function like a smart clerk, pinpointing clerical errors and completing forms with lightning speed, LegalSifter is more like that 80-year old senior partner or old solo who’s been practicing forever, who can glance at an agreement for thirty seconds and identify at least three issues that never occurred to you.

So does Legal Sifter mean the end for lawyers? Hardly. For starters, we lawyers can use LegalSifter ourselves to quickly evaluate contracts for clients. I also see Legal Sifter as a great training tool for new lawyers who haven’t yet acquired eyeball skills – so long as they don’t blindly rely on the LegalSifter results but instead, use them as a basis for further research.

Nor do I see LegalSifter, used as a stand alone tool by the end-user client as being harmful. Today’s clients already have dozens of ways to get in trouble online, including plucking a form off the Internet that may have limited applicability to or benefits for their situation. At least LegalSifter highlights some of the potential problems. At the same time, because LegalSifter is a machine, it’s unlikely to create a false sense of security as most users are smart enough to realize its limits (just as I argued that most consumers are unlikely to go out and sign up for chemo treatments based on the results of a home DNA kit).

Where I do see the potential for harm, however is the gray area that lies between the traditional practice of law by licensed attorneys and the computer-generated analysis conducted by machines on the other. It’s one thing for clients to use LegalSifter directly because as I explained, they understand the limitations. It’s quite another for banks or grocery stores or other non-lawyers to set up a LegalSifter kiosk manned by non-legal support who simply run clients’ data through the machine and then explain the results generated or offer additional advice without necessarily understanding all of the legal issues that may be involved. In any event, because LegalSifter enables lawyers to operate more efficiently by churning through a first level of contract review, there’s no reason that actual lawyers couldn’t set up a similar operation to compete on cost but offer the added benefit of review by a real lawyer, along with the added protections of privilege and malpractice insurance that non-lawyers can’t provide.

Even if lawyers aren’t ready to embrace LegalSifter, the worst thing we can do is ignore it.  Give LegalShifter a test run so you can understand what it can do. Because there’s no doubt in my mind that LegalSifter is a game shifter.


  1. Paul Spitz on September 2, 2014 at 11:30 am

    Are you sure that clients understand the limitations? I’m not.

  2. Paul Spitz on September 2, 2014 at 12:00 pm

    Heh, I broke it. I uploaded a fairly complex stock purchase agreement prepared for venture capital transactions. After a minute or so, Legalsifter admitted defeat. But the messages as it was trying to “analyze” the contract were quite amusing.

  3. myshingle on September 2, 2014 at 12:04 pm

    Seems you won’t be displaced any time soon! Clearly, this tool is not yet ready for complex agreements – and even if the technology emerges, it is probably not cost effective to develop a tool unless there are enough of those types of Ks to justify it.

  4. Paul Spitz on September 2, 2014 at 10:15 pm

    Well, that’s the problem with these “disruptive tools.” Legal services ARE complex. There are a lot of issues that take not just education, but experience, to understand. That’s why we go to law school for 3 years. That’s why we practice for several years before we ever feel confident.

    Some algorithm based on key words devised by an engineer who has not spent 1 day in law school, and couldn’t read 1 paragraph of Henry James with any level of understanding, cannot come close to replacing what we do. If a client thinks that googlesifter is going to protect him in even a run of the mill transaction, well, a fool and his money are lucky to get together in the first place.

  5. Paul Spitz on September 2, 2014 at 10:34 pm

    Continuing with my thought, anything based on key words is inherently flawed, because contract is as liability and contract and warranty contract and liability and liquidated damages. Wow, I just used a whole lot of keywords, and it makes no sense. Any unemployed JD who has done document review with these doc review software programs know that the ratio of false-positives to actual relevance is something like 1,000,000/1. But you have to wade through a million e-mail chains manually to weed out the crap.

    So, Stanford computer science grads should stay the frack out of my profession, and I’ll stay out of theirs.

  6. Mountain Sports on September 29, 2014 at 3:04 pm

    Nice post and fyi — it’s ShakeLAW.com

Leave a Comment