Is Contract Review Dead?

Thoughts on what might click, and what won't

I had earlier written a brief post about what will differentiate lawyers in a post-AI world.

My thoughts have evolved over the past few months and I thought it is time to share them.

What’s Happening in the AI World?

Over the past few months, LLMs have started performing far better than they used to. The differences are perceptible especially when you consider how new LLMs are performing over longer horizons of text, which makes them more effective than before at review tasks.

We can go so far as to say that review and analysis which were the moated fortresses of legal (and the wider consulting) profession are been violently breached as we speak.

How?

One year ago, when we built contract review tools, anything over 1000 words needed to be broken down into smaller pieces and then analysed separately, and then stitched together. This made some errors creep in.

And then because one part is being reviewed without looking at another - scope for errors, misinterpretations and lack of context compounded the errors.

All this meant automating review was complicated enough for us to spend considerable amount of time trying to make work.

Now LLMs are delivering highly accurate review of documents with length even in excess of 15,000 words.

Effectiveness of reasoning models also means that analysis that flows from this review is of higher quality than we imagined.

Where Does That Leave Those Who Review Contracts?

Knowing that everyone has access to the models I have, how would I re-imagine value-add in contract review tasks today?

First, we should all recognise the benchmark. Your business team or client is one-shot prompting with a contract and getting fairly useful stuff out of that.

Second, as I wrote in the earlier post - people come to you because they trust you.

At the same time, they expect more and in less time because they know you atleast have access to the tools you have.

Client Expectations

This means expectations are that your contract review must be:

(a) qualitatively superior, layering your real world experience and insights;

(b) designed for your audience, whom only you understand and not your language model; and

(c) tasteful, it must carry your signature, flavour and style.

What About AI Powered Contract Review?

Once we were convinced of the quality, accuracy and comprehensiveness of the review we were able to achieve, when building our contract review web app, we spent an inordinate amount of time in figuring out how the results should be presented.

There can be different opinions about this and that’s why I believe it is a matter of taste.

Also, we are conscious that it is meant to be a general use tool for contract review, and particular type of contracts might be possible to present in a different manner.

So - snapshot at the top, key details below that. Then we move on to the substance of the analysis - first by running through all key terms and their analysis if the user wants to get into so much detail.

Recognising that many users might just want to jump straight to the risks and recommendations, the detailed analysis only expands if you click on it.

The meat of the analysis is in the risks and recommendations. Here, there might still be users who skip straight to the recommendations - and that’s okay too!

The recommendations are structured in as much of an actionable manner as possible, and the overall language of the report is meant to be understood without lawyer-level reading comprehension.

This report is downloadable as a PDF that I believe looks nice.

It has to look nice.

This is something lawyers new and old have always obsessed over, and for good reason.

The matter of taste, the matter of your audience and the matter of being someone who can be trusted, shorn of all other trappings and pretences will be the driving traits of future legal professionals.

Times of churn are times of great opportunity.

I expect that those who will thrive will obsess about these aspects with even more fervour, their zeal reflected in the AI that does their work, where they direct not only what must be written but how it must be written, and conscious of the needs and expectations of their audience.

What a time to be alive!