(Credit: 3dshtamp/Shutterstock.com) (Credit: 3dshtamp/Shutterstock.com)

 

It’s sometimes difficult to be discerning in purchasing and using new and largely unknown technology. Underlying some of these technologies is early stage artificial intelligence, and some legal departments are using it for contract management. But while much has been said of the benefits of AI-enabled contract management, little is known about how best to purchase, maintain and implement such tools.

At the “Using AI to Support Contract Management: Dirty Secrets and Best Practices” session at the 2017 Corporate Legal Operations Consortium’s annual institute in Las Vegas, legal operations managers from NetApp and Microsoft sought to change that. Both company’s legal departments are partnering with vendors on implementation of their AI contract technology.

During the session, they shared their advice on how to best to implement and use contract management solutions without getting lost in the unforeseen details.

1. Train AI with a Large, Diverse Data Set

AI contract solutions essentially “learn” / become familiar with the types of contracts created and managed by a legal department, and this takes a lot of time. And perhaps surprisingly, the process is simplified when there’s an abundance of information available. For when the technology reviews smaller data sets, it “can’t find a lot of common denominators,” said Tami Baddeley, operations lead at Microsoft.

AI contract platforms need to find such “denominators” to tell whether certain contracts or clauses deviate greatly from the “standard” contracts legal departments use for specific subjects or litigation areas. Ginger Dolgow, senior manager, Global Legal Shared Services at NetApp, said one should expect to run “tens of thousands” of templates to achieve a suitable level of accuracy.

She also advised training the contract solutions on unique contract types, such as “third party papers,” and “a most favored nation clause,” so it can become familiar with the different deviations it can expect to find.

2. Don’t Expect Automation to be Automatic

With AI contract management, it is easy to think machines can automatically draft, review approve or reject contracts completely on their own. But that is far from the case. “There’s no magic button. We all look for one, we all want one, but it doesn’t exist,” Baddeley said.

She added that because AI will not always be entirely accurate, “most of the systems require some type of involvement from people in your team’ to review [the solution’s] work.”

Further, AI contract systems require upkeep and maintenance. “Don’t think you’re going to hire a company, they are going to solve your [contract] problems and after they’ll go away,” Baddeley said, advising legal departments to retain their vendors. Another option is to rely on internal or outsourced IT staff on an ongoing basis to help upgrade and fix a system throughout its operational life.

 

3. Ask About Hidden Pricing Upfront

Ongoing maintenance also means that AI contract solutions will cost far more than their sticker price.

“You need to know what types of resources you need, and you should ask [vendors] up front, ‘What is the total cost?’” Dolgow said. “[Tell them to] give you everything you need to consider, because often what you are presented with is only the out-of-the-box cost for the software itself.”

Some of these extra hidden costs, she added, can include everything from training the technology to hosting fees and minimum spend requirements. Companies should also take the time to consider what pricing model is right for their implementation and need.

“There is a variety out there, one size is not going to fit all,” Dolgow said.

4. Keep your Metadata, or Be Stuck with Your Vendor

When using AI to review contracts, legal departments can pull metadata from across their documents to understand a variety of variables, such as what types of contracts they hold, when their contracts were approved and stored, and what liability or protections their contracts contain.

But often times, this metadata is collected and stored by the AI vendor.

“When you cull that information out of whatever system you have, it can be [connected] with your supplier,” warned Baddeley. Such an instance would mean that a legal department would have to rely on the vendor to access information about their contracts.

This has driven legal departments at Microsoft and NetApp to make it a point to store their own contract metadata in-house. “I would suggest it, because then you’re are not dependent on the vendor’s data set,” Dolgow said.