Automating Contract Creation, Without Help from Legal: Test-Driving LISA (Part 1)

If you thought your cozy, knowledge-based desk job was safe from the insatiable beast of automation, we have some bad news for you, procurement people.

Advances in artificial intelligence (AI) are starting to imperil professions based on more complex mental capabilities. Take, for example, how Japanese insurance company Fukoku Mutual Life Insurance is reportedly replacing 34 human insurance claim workers with “IBM Watson Explorer,” starting by January 2017.

The AI will scan hospital records and other documents to determine insurance payouts, according to a company press release, factoring in injuries, patient medical histories and procedures administered.

We don’t mean to send you into a panic (especially right before the weekend starts). But it’s an uncomfortable reality that many procurement tasks are ripe for the same kind of automation.

Luckily, insurance underwriters and procurement clerks aren’t the only workers at risk. One new technology we recently tested may actually give supply managers a new tool to use while making someone in the legal department redundant.

Meet LISA

In the midst of our ongoing research on how AI is changing contract management, we encountered an intriguing tool that allows users to create an NDA from scratch — without a lawyer.

Created by AI Tech Support Ltd. and powered by Neota Logic, LISA (Legal Intelligence Support Assistant) “has been programmed to help you and ‘the other side’ (the receiver) find a commercially sensible middle ground for your NDA.” Instead of the usual back and forth over small points, led by biased human lawyers, LISA allows you hash out an agreement fairly quickly (10–15 minutes) through a web app.

What can LISA do? According to the site, LISA will help you and your intended receiver to:

  • Agree on the substance and draft a bespoke NDA so that you have a final document you are both comfortable in signing and honouring
  • Understand the key legal and commercial principles behind your NDA that you need to consider
  • Find a middle ground for you and your receiver through assistance (guidance notes) and prompts via pop up boxes to help you both move forward
  • Consider if you want to move beyond an NDA. For example, you may wish to consider a memorandum of understanding, shareholders’ agreement, joint venture agreement, etc.

Our Top Secret NDA

As editors, naturally our skepticism kicked in. Surely it couldn’t be this simple. So we decided to create an NDA ourselves.

The process is straightforward. Users are asked to submit the usual details (name, email, etc.) and then are guided through a series of web forms with questions to complete.

Sample questions or section headers include:

  • How do you want to define "representative"?
  • What information do you want to protect?
  • Are additional security measures required?
  • How long will this NDA last?

There are also several opportunities to include detailed language on the nature of the agreement via standard web-form fields, for example, “particular confidential information” for specific inclusion (outside the boilerplate).

At the end of the process, an NDA appears for you to review. After you scrutinize LISA’s work, you simply click “Email to Receiver” and off goes your NDA to the counterparty.  

Apparently this is all it takes to create a legally binding document. It’s all very easy to do. Almost too easy.

One of the first questions that popped into our heads was this: “Is this really AI?” While the system obviously works as intended, the logic behind the program, from the outside looking in, seems to rely on web forms and conditional logic.

Helpful and efficient? Absolutely. True artificial intelligence? Depends on the definition.

Granted, we have neither a computer science background nor have we actually gotten a peek under the hood at LISA. So how does it actually work? Stay tuned for Part 2 to find out!   

Share on Procurious

Discuss this:

Your email address will not be published. Required fields are marked *