ironclad logo

Train Your AI Model With Custom AI Clauses

By subscribing, you agree to receive other communications

Read the transcript

Out of the box, Ironclad analyzes your incoming contracts for 194 different terms and clauses, like counterparty info, payment terms, and termination rights.

But what if you need insights and guardrails for positions that aren’t included in these out-of-the-box properties?

Well, now you can train your own deep learning models with a feature called Custom AI Clauses.

To see how it works, let’s say “customer data” is a clause we want to add to the system.

We’ll copy it and head over to the new AI Clauses tab. And here you can see we’ve already begun training the custom model with two different customer data clauses that we’ve pulled from other vendor agreements.

And so we’ll add this one to make three total examples, and three is all we need. There’s no coding or large data sets required. We enable it and it’s ready to go.

And from this point on your repository imports will flag these clauses automatically. Beyond that, you can begin to build compliance guardrails around these properties too.

Let’s take a look at it over in Workflow Designer, where we’ll select customer data from our AI Clauses. And then require legal approval if the clause doesn’t appear.

Okay, let’s test it out. Our stakeholder brings in a new vendor agreement. Ironclad AI analyzes it, and here it is, a customer data clause. This one is written differently than the examples we trained on, but Ironclad AI identifies it nonetheless, and as a result, no legal intervention is needed at this time.

And that is Custom AI Clauses available in beta, starting in April 2023.