If you’re looking for a contracts technology that’s at the opposite end of the enthusiasm spectrum from AI, I nominate contract automation, also known as document assembly. But it offers the only way for us to escape copy-and-paste heck. In this post, I explain why and how.
The Challenges
Contract automation allows you to create a contract by completing an interview. You consult the guidance, then answer the questions. When you’re done, you click a button and the system pulls together preloaded language and delivers a draft contract that expresses the deal in a way that is consistent with the choices you’ve made.
The contracts community has looked sidelong at contract automation and responded with a collective Gallic shrug. That’s because contract automation poses challenges the contracts community finds it difficult to tackle. (The discussion of challenges is derived from this 2023 blog post.)
First, it requires figuring out what to say in a contract and how to say it, taking into account the customization you want to offer and how all the different scenarios interact. If all you’ve ever done is ride the copy-and-paste train, that sort of work would be scary.
Second, you have to do a bunch of coding. Your contract language ends up swimming in a sea of code.
Third, there’s the issue of whether you’re able to reap economies of scale, with many using a given template. An organization considering its own contract-automation project might well decide that the use it would make of the template wouldn’t justify the cost.
And fourth, there’s inertia. Any kind of change will, in the short term, impose costs that are greater than the benefits. Changing your contracts forces everyone who touches them to adjust their thinking. If you’re wedded to short-term thinking, it’s easy to reject the change that comes with contract automation.
All the Questions
I’ll now add a fifth challenge: getting those who work with contracts to tolerate answering a bunch of questions the first time they tackle the interview for a given template.
The questions are a feature, not a bug. They’re the result of the user being presented with a broad decision tree, leading to a more relevant contract. But if all you know is copy-and-pasting, you’re used to being presented with an attenuated decision tree. You’ve accumulated cognitive debt: in other words, you’ve acquired the habit of being fed answers without knowing how they’ve been arrived at. Being presented with an interview that requires you to make choices can come as a rude shock. (I wrote about the too-many-questions issue in this 2023 blog post.)
My Failures
I’ve long seen contract automation as the only way to achieve a fresh start for contracts. I made my first attempt around 15 years ago, with my friends at ContractExpress (before it was acquired by Thomson Reuters). I created an automated confidentiality agreement and set up my little lemonade stand. And nobody came.
Around three years ago, I got a second chance. The then CEO of LegalSifter, Kevin Miller, took pity on me and allowed me to create Adams Contracts, a division of LegalSifter devoted to creating highly customizable automated templates. So I’ve built three templates, for creating confidentiality agreements, boilerplate, and straightforward service agreements. The templates are great. I got up on my soapbox again, but hardly anybody came.
Doing Better
In the article published today by Corporate Counsel Now (go here for the article; go here for my related blog post), I again say that contract automation is the solution. Why wouldn’t that be a washout too?
To address the challenges posed by contract automation, what’s required is what has been obviously missing in my two attempts. First, sufficient resources to allow the building of a template library with a critical mass. Second, the involvement of different organizations active in contracts. And third, a concerted attempt to win over the relevant constituencies—something more sophisticated than my nagging people.
This Isn’t a Nail for Your AI Hammer
But why contract automation? Because you use for a given job the tool the job requires. I know many of you are itching to use your AI hammer, but fixing dysfunctional contracts isn’t a nail. AI could have an ancillary role, and AI would benefit in the long run, but the tool for fixing contracts is contract automation.
Because today’s Corporate Counsel Now article involves a merger agreement, I use the article to suggest building a library of highly customizable M&A templates. One has to start somewhere, so why not M&A? It would seem to make sense to aim to serve mid- to small-market M&A, where inertia perhaps has less of a stranglehold.
Chaos Offers an Opportunity
It seems that many in contracts are thinking in terms of segueing from copy-and-pasting to AI. The result would be moving from one form of dysfunction to another, with users perhaps acquiring yet more cognitive debt.
But the current confusion offers an opportunity: while things are in flux, let’s insert a broad-based contract-automation initiative into the system. We’d all benefit.
What meaningfully distinguishes copy-and-pasting from having a computer knit together pre-written snippets and variants, apart from efficiency? Could a law firm have implemented the better way you have in mind entirely by paper process in 1950, with typing conventions and a secretary pool?
Hi Kyle. Two things. First, most organizations can’t expect to compile optimal contract language: after endless copy-and-pasting, most people are to riddled with cognitive debt. So avail yourself of an authoritative automated template that really is reliable.
And second, a highly customizable template allow for customization way more effectively than does an improvised approach: you decide what you need, and the template serves up language that’s clear, concise, and relevant.
Ken