Customising legal workflows using AI

Some thoughts on how to approach adaptation.

How do you customise AI for legal workflows?

A few lessons from helping professional teams in their AI transformation which we believe should guide legal technologists.

1. Understand the subject matter.

Begin by understanding the legal / professional expectations, problem statements, workflows and challenges. This is more than 50% of the work!

How information is gathered from a client, how a draft is prepared and reviewed. What templates are used, what variations are seen. How is research undertaken. What documents are referred to. How much time is spent, who spends it, and how they spend it. This helps figure out what are the painful parts, what is time consuming and where there is scope for improvement.

2. Choose low hanging fruit.

In any workflow, it is good to start with things that are easier, before moving over to more ambitious parts. There are things AI does very well, and things that are iffy, or difficult. There are also things that don't require AI but can be solved using simpler methods. Bringing these ideas together helps achieve returns on investment earlier rather than later, and build a base to target more complex parts of the workflow. Let your work put $ on the table.

3. Elevate the weakest, enable the strongest.

Applying AI to workflows means recognising different challenges exist across an organisation. Younger professionals need help with quality and consistency to produce better first drafts, senior professionals might benefit from easier reviews, revisions and research challenges.

Any AI transformation has to work to remove bottlenecks and empower people where ever they are, whoever they are.

It should deliver observable results. For example, you should not expect your clients to pay for far down the road experiments such as model training exercises to nowhere.

The future is being won by technologists and professionals who are able to understand each other and execute.

Never been more interesting times!

Would love to hear your experience. Thanks for reading.