How poorly drafted clauses lead to litigation: Q&A with Maryam Salehijam, PhD

Maryam Salehijam’s background is in European law and international law. She went to Maastricht University for her LLB, studying European law, followed by an LLM in international law. She then completed her PhD in International Business Law with a focus on commercial dispute resolution at Ghent University. She is now on the GTM team at Robin AI. She is a guest on the latest episode of The AI Sidekick, with host Ryan Heath.

Listen to the episode here:

Spotify

YouTube

Apple Podcast

‍

What’s biggest problem with contracts today?

Companies don’t invest in quality contracting and then they pay the price. Contracts are seen as that very important, invisible engine behind a business, but most contracts are drafted really poorly, they’re sloppy. Then at some point something goes wrong. For example, a customer isn't paying, or someone is claiming that service wasn't delivered, or there's some big event like COVID-19, and everybody runs back to the contract to see what it says. And because they're drafted so sloppily, the contract can be very inconsistent. It can say one thing in one place and another thing in another place. Then it turns into a dispute of what the contract actually says, and businesses end up spending millions of dollars disputing what their contracts mean.  

As a result, the biggest expense for businesses, when it comes to legal spend, is litigation. But if companies could have better contracts, they wouldn’t have so much litigation.  

What are the primary challenges that lawyers and general counsel encounter when drafting complex contract clauses?

It all comes down to the regulatory landscape. Privacy is different in every state and every country, so it constantly changes depending on where your client is based. What ends up happening is that anytime there's a regulatory change, legal teams have to rush back to their contracts to see if they need updating.  

We saw that with the California Consumer Privacy Act, and everybody needed to update their Data Protection Agreements, and ended up costing businesses millions of dollars because they had to throw humans in to review the DPAs.  

How do you envision AI transforming the drafting process of complex contract clauses?

Tightening the language and making it consistent. Today, a lot of contracts are vague in their language. For my PhD, I went through 172 contracts line-by-line to see where their inconsistencies lay, and it took me two years to analyse that.ⁱ Now, Robin AI can do that in minutes. That’s fascinating, because it means that as a business you can very quickly see where the variations in your contracts are, look at the language and see what’s specific and what’s too vague, and clean them up to make sure they’re consistent. Right now, businesses just don’t have the bandwidth to do that with human labor, which is why generative AI can really change the risk profile for legal teams.  

How often do poorly drafted alternative dispute resolution (ADR) clauses lead to litigation?

Since the ‘80s, most commercial contracts have described how a conflict that arises from the contract ought to be resolved. Why? Because companies realized that if you don’t prescribe how you’re going to fight, your opponent might fight dirty, and take you to courts you don’t want to be in. So they started prescribing that if there’s a dispute, for example, we’ll go to the courts of New York, and the contract is governed by the laws of Delaware.  

That worked, but companies quickly realized the court system is slow and expensive, so they needed an alternative. Arbitration came with the promise of being faster, cheaper and better. But in practice that wasn’t always the case. So mediation or negotiation got introduced as the step before that.  

The typical dispute resolution clause is terrible: it’ll say they’ll mediate or negotiate before arbitration, and if that doesn’t work, they’ll go to court. But they really don’t explain how long, who, where, what, why. And so most litigation starts with a dispute regarding jurisdiction and forum, meaning companies can spend months, if not years, fighting about how they’re going to fight.  

How does the speed offered by AI overall benefit contract lifecycle management (CLM)?

A lot of companies bought CLMs in 2015-2024, because they promised the ability to know exactly what’s in your contracts and where to find it. However, CLMs require a human to go in and manually link everything. This is why it can take years to implement a CLM and you still can’t search well across them because, unless you're already told that what you need to look for as the contract went in, it's not very good at going back and looking at things, and that's because most CLMs are not built on native Gen AI. But with the advent of Gen AI and native AI companies, it’s now possible to solve that problem.  

The future is going to include partnerships between experienced AI providers and CLM providers – so that you have a repository and drafting and editing tool in one place, or the AI companies taking more of the market that CLM providers thought they owned.  

Can you provide an example of a clause that AI has significantly improved in terms of clarity or enforceability?

Something that AI does really well is confidentiality clauses. A lot of times it's one sided, and it's so easy to make it mutual. You always want privacy, confidentiality, and non-disparaging clauses to be mutual, and they rarely are to start with. AI is so good at that and it takes it two seconds to do.  

What would your advice be to a General Counsel starting out on their AI journey?

Have realistic expectations. AI is much better at amplifying your human talent than replacing it.  But you should incorporate AI now, because you don’t want to be starting from scratch when a next revolutionary leap forward happens with AI.  

‍