Imagine you are representing a client in multiple matters related to a number of similar patents. You’ve finished reviewing documents in the first few matters, and have identified which ones are privileged, which ones are junk, and which ones are likely relevant to specific issues. Then another matter is started with a different party. You end up having to collect from even more custodians and load new documents into a database. Where do you start?
What if there was a way to make sense of 20%, 40%, or even 60% of these documents based on the work you’ve already done? What if you could quickly identify which documents are likely to be relevant to specific issues like invalidity or infringement? What if you automatically sort documents by type (e.g., “scholastic journal” or “code”)? And what if you could easily eliminate all of the junk you collected from your new custodians and all the junk that was produced to you?
With this kind of head start, you’d be able to set your legal strategy earlier. You’d be able to engage the right witnesses sooner and write better discovery motions. Not to mention you’d be able to save your client tons of money on review.
With the technology that exists today, there is no reason the artificial intelligence (AI) you use in document review should have to relearn everything from scratch in every new case. After all, you don’t have to relearn how to ride a bike or drive a car every time you get a new one. You remember what you’ve learned and get going immediately.
So, we built a new kind of AI at DISCO.
Introducing DISCO cross-matter AI
We are excited to announce the launch of DISCO cross-matter AI, which ensures you never have to start a review from scratch again. DISCO AI remembers what kinds of documents were relevant to different issues in your matters, and immediately offers you more documents like them as soon as you spin up a new case. This allows you to find the documents you are looking for faster and conduct your reviews more efficiently — which will result in better legal strategy earlier and better outcomes.
How does DISCO AI work?
DISCO AI models provide tag predictions that help you find the documents you are looking for faster. DISCO AI observes in the background as you tag documents, and learns to predict which types of documents are likely to receive certain tags. As you review more documents and agree or disagree with the predicted tags, DISCO AI continues to observe, re-train, and get smarter. It’s a lot like recommendations on Netflix™, where your recommendations get better and better as your watch and rate more movies.
Cross-matter AI extends the impact of your DISCO AI models. You can now build master models made up of any combination of DISCO AI models from your individual cases. You can link your cross-matter AI models to your new cases to start receiving tag predictions immediately. You can also choose to send the tagging signals from your new case back into the cross-matter AI model, ensuring your models continually get smarter. This bi-directional learning is an industry first.
Cross-matter AI across all of your cases
While cross-matter AI will have a huge impact on cases with similar issues and millions of documents, it can also help with every case on your docket, no matter the size.
Every practice area has similar document types across cases — whether it’s scholastic journals in patent cases, clinical trial records in pharmaceutical intellectual property disputes, or communications about repairs in construction cases. With cross-matter AI, you can easily identify documents like these as soon as they are loaded into DISCO Ediscovery.
Then there is privilege, where making a single mistake can have huge repercussions down the road. With cross-matter AI, you can build a high-accuracy model based on hundreds of thousands of signals from across your matter, and use it to identify any potentially privileged documents you don’t catch through search terms or manual review. You can QC these documents before your production goes out the door, ensuring no privileged documents slip through the cracks.
Cross-matter AI is a game-changer for law firms
Cross-matter AI will change the way law firms practice. Not only will they be able to save their clients money with more efficient reviews, but getting up to speed faster will mean better legal strategy from the start. Pleadings will be more specific, discovery negotiations will be better informed, and witnesses can be engaged earlier. And more disputes will be resolved before lawsuits are even filed.
Additionally, rather than throwing away all of the judgment calls they make during their cases, law firms will be able to productize their expertise. They can create custom models for issues they see repeatedly in their practice areas and market these models to their clients.
Cross-matter AI is a game-changer for corporations
Cross-matter AI will take a huge burden off corporate legal departments, which are already trying to stretch their finite resources. Corporations can create their own models for routine internal investigations — like discrimination or inappropriate workplace conduct — and their in-house teams can use them to quickly determine which cases have merit and how to resolve them.
Cross-matter AI will also allow corporations to maximize their investment in outside counsel across the board. Rather than paying for protracted document reviews to weed out junk or find privileged documents, they can focus their spend on high-value legal work early in their cases. Not only will this save money, but it will allow cases to be wrapped up sooner — which will free up legal teams and internal employees from the demands of litigation.
All the benefits, none of the risks
Our technology ensures that you can share cross-matter AI models throughout your organization without creating any risk of the underlying data being shared. Unlike traditional support vector machine (SVM) models or “bag of words” algorithms, DISCO AI only uses abstract representations of documents and learns from these representations, and doesn’t record which ones it has seen. This means there is no way to reverse engineer the underlying data based on the model. As a result, you can start using cross-matter AI today with confidence.