With Dioptra Wilson Sonsini Launches Super-Accurate AI Doc Review

Wilson Sonsini’s Head of Innovation, David Wang, sums it up: ‘You cannot have efficiency without accuracy.’ And that has always been a challenge for legal AI doc review. Now, in what Wang sees as a breakthrough, the leading California-based law firm has brought aboard Dioptra, a stealth mode legal AI startup, which through the firm’s Neuron group is hitting extremely high accuracy on contract review. The AI company states it is reaching 92% consistent accuracy on third-party documents, for example.

‘This is a Sputnik moment,’ Wang added, referring to the time when a human first orbited the Earth and started the space race.

Now, you may say: ‘Hold on, I thought some legal AI systems were already 99% accurate.’ Well, the reality is that, at least from Wilson Sonsini’s point of view, there is a difference between ‘demo accuracy’ and really getting it right on very complex contracts where the work needed involves a deep understanding as well as subtle changes to the language, which may in turn demand the alteration of other aspects of the agreement. It’s what Wang calls ‘dimensionality’ and contract review has a very high dimensionality level.

I.e. this is not something as basic as an AI system finding a clause or not finding it, or even directly matching a sample of a clause’s text to that held in a playbook.

The end result of the AI part of the process here is something that a partner would accept on the basis that a human lawyer then made some small changes, Wang noted. But, beyond those small changes they’d stand behind it, just as they’d stand behind any other work the firm did for a client. That is how confident they are of the results. And in this case, the contract review type now offered by Neuron is for cloud services agreements.

‘Partner assessment of a piece of work is the hardest test there is,’ Wang added, ‘especially when there are sections of a contract that may have 20 different aspects to them, which if changed alter other parts of the agreement.’

The AI is also very adaptive in how it works, or as the company states: ‘Dioptra does not rip and replace [language when doing a contract review]. It minimizes redlining and leverages existing language.’

That is to say, it operates more in the way a human does when editing a piece of text, in that it seeks to understand what’s written there and then works with that to get where it needs to be. A human editor doesn’t just drag and drop batches of words from one place to another, for example, they understand the intent of the writer and then tweak it so it’s correct.

The offering, as part of the firm’s specialist Neuron group that focuses on digital solutions for startups, will be provided on a fixed fee, and as Wang stated, beyond the excitement over the technology ‘this is about client service and helping them get to revenue faster’.

So, on one level this is simply doing what clients want: solving their legal needs – quickly. On another level it’s doing something that is fundamental to the eventual widescale roll-out of legal AI across the sector: providing such a level of accuracy that efficiency can really take hold. And when there is that level of confidence, things can change a lot faster than in the past.

David Wang, Wilson Sonsini

As Wang has argued convincingly, legal tech companies can talk as much as they like about helping to reduce inefficiency, but if the results of those tools are not accurate enough to give partners confidence, especially for the kind of ‘multi-dimensional’, highly complex, high-risk contracts that large commercial law firms handle, then they cannot really deliver what is needed and end up triggering so much follow-on work that some of the efficiencies are lost. (Or the partners just don’t use those tools for such sensitive matters…)

To reiterate the central point, as Wang states: efficiency depends on accuracy, because without accuracy the entire AI-enabled process breaks down, or at best is partially self-defeating.

The goal now is to keep pushing up the accuracy level and to also move onto other contract types. Wilson Sonsini stated they are also exploring the same offering for: ‘Terms of service, non-disclosure agreements, and agreements for other verticals, with a focus on those that unlock revenue and enhance the client experience in the provision of the firm’s legal services.’

Plus, they added: ‘With over 40% of in-house counsel at technology companies stating that contract-related tasks occupy more than half of their day, contract review is a major pain point for clients.’

And let’s just repeat that this offering is on a fixed fee. So, predictable costs and reliable accuracy. Now that is the path to achieving real legal efficiency.

Clearly this is just the start, but it points the way. How this evolves will be fascinating to watch.

And if you found this inspiring, then come along to Legal Innovators California conference, June 4 + 5 – San Francisco, where David Wang will be giving a keynote on some of the above themes, such as accuracy, evaluation, and how AI can really drive efficiency.

Legal Innovators California conference, Jun 4 + 5. 

The event will take place in San Francisco with Day One focused on law firms, and Day Two on inhouse and legal ops. We have many great speakers, along with a group of pioneering legal tech companies and service providers – you can see some more about our speakers here. It will be two great days of education and inspiration! Join us! 

For ticket information, please see here. Don’t miss out on what will be a great event in the heart of America’s tech world.