As pioneers of AI-driven contract review and creators of the Contract Review Automation (CRA) category itself, we are always interested in seeing new solutions come to market and their approach to automated contract review. While all CRA solutions leverage varying degrees of AI, we have divided these providers into two categories: Self-Service AI models (i.e., machine-only) vs. Supervised AI platforms that combine human intelligence with AI technology (i.e., LawGeex). 


Quality is Table Stakes

As a company made up of experienced attorneys and data scientists, and after several years providing solutions for lawyers, we can say that quality is by far the most important criteria. Contracts are complex and sensitive in nature, and lawyers are especially cautious when it comes to contract work. One small error or oversight can have detrimental effects on a company. 


Our experience has taught us that AI-alone is not yet sophisticated enough to handle the complex nature of contracts. This is supported by several legal teams we spoke to who say that Supervised AI solutions are more widely adopted because–in addition to returning fast and accurate results–they provide “peace of mind” to truly delegate contract work, saving a significant amount of review time that can then be spent on higher value work. 


Legal teams view Self-Service AI solutions as “assistant-only” tools because the first-pass results still require review. Just like the difference between a great paralegal and a mediocre one, one actually saves you time, the other creates more overhead. 


It all comes down to accuracy. When the output is not accurate, the legal professional is required to read the contract from A to Z, fix inaccuracies, and even worse–repair legal mistakes. 


The Unsupervised CRA Blindspot 

A major blindspot of Self-Service AI is that it learns from client contracts alone. If the samples provided by the client do not include a specific legal issue (for example, because it’s very rare – think about an IP Assignment clause), Self-Service AI will not be able to identify them. 


Besides providing higher quality, Supervised AI review enables the system to learn more from the supervising humans who can perform a holistic analysis for intent of the contract terms, where Self-Service AI-only review is completely literal. In short, a Self-Service AI solution can be a great assistant tool to the legal team, but only in the shape of a junior-level resource which still requires significant oversight from the in-house team. Legal teams will still need to read the whole contract to mitigate risk and ensure there aren’t any red flags hiding in the contract.



In addition to being far more accurate, Supervised AI solutions are also easier to deploy. They are often “pre-trained” which means the AI has already been taught to identify and understand the various legal concepts and legal nuances using external legal data. In the case of Self-Service AI solutions, the machine needs to process significant amounts of client-specific data in order to achieve reasonable results. This creates an onboarding process that can take up to a few months vs. a few days/weeks for Supervised AI.



When evaluating speed, Self-Service AI solutions definitely return a faster output. However, it is even more important to look at the overall time of the contract journey. Since the Self-Service AI solutions function as “assistants” to the legal team, the client is still very much dependent on the speed and availability of the legal team itself. In other words, it is still reliant upon the legal team to perform the review. 


It is critical to take into account the entire lapse of time between the moment the contract is being sent to the AI solution and the moment the contract is actually ready for signature. 


Supervised AI solutions return superior results from the onset, so the business is not dependent on the legal team’s review, resulting in a much faster overall time to close the deal. This also means other business units can tap into the power of the Supervised AI CRA solution directly to easily send contracts for review and receive redlined contracts back based on the company’s policy, essentially eliminating any legal bottlenecks.


Similarly, when the redlined contract goes back to the legal team from the Supervised AI, the second pass review happens much faster due to the higher accuracy levels. This results in significantly higher time/cost savings for the in-house legal team.

AI vs Lawyer

LawGeex was actually featured in HBO’s “The Future of Work” series in an episode that pitted our AI against an experienced corporate attorney. Spoiler alert: our AI outperformed the human lawyer. The attorney took over an hour to review two contracts which he did with 85% and 83% accuracy, while our AI achieved 95% accuracy on both contracts within 18 minutes. 

And while it’s true in this case that our AI-alone solution outperformed the attorney, what we have heard from CRA clients and know to be true ourselves, is that the combination of AI + human intelligence (Supervised AI) offers superior accuracy and is far more reliable than unsupervised solutions (Self-Service AI) alone.



AI alone can not yet achieve the flawless precision legal demands. AI as a technology hasn’t fully matured yet and still needs to be trained. Do you want to train algorithms and supervise AI, or do you want to free up your schedule to work on higher impact work? I think we both know Supervised AI is the way to go. 


Unsupervised Solution’s output


LawGeex’s output


To learn more about AI for contract review, download our free eBook or get in touch with us 

The post Not All Contract Review Automation Solutions Are Created Equal appeared first on LawGeex.