In the vending machine analogy, the client receives none of the nuance or learned experience that a lawyer can apply to their situation.
Legal AI has been rapidly adopted. Most law firms now use a legal AI tool, and most LLMs now offer some sort of legal function. There’s a clear expectation of legal AI: prompt in, product out. Now, however, the legal industry has a ‘vending machine’ problem. Thanks to the reach of legal AI, there’s now a vending machine inside every firm, which means the product remains largely constant. As a result, clients could well receive the same product everywhere.
To follow that thought – if all law firms use the same underlying models to produce the same work as their competitors, two problems will crop up: what is left to differentiate one firm’s expertise from another? And crucially, how do you stop any practice masquerading as a Magic Circle firm?
Expertise is still the product
The law firm’s product is, has always been, and always will be, judgement. A client rents the law firm’s expertise in order to leverage the skill and judgement acquired through years of experience. AI changes nothing about this requirement, and in-house experience still defines a law firm’s competitive edge.
In the vending machine analogy, the client receives none of the nuance or learned experience that a lawyer can apply to their situation.
Increased output without differentiation wins nothing
Generalist legal operating systems – which now include Claude and OpenAI – are brilliant at enabling a law firm to amplify output within their practice. However, while this enables firms to do more work for a client, it’s now imperative to make sure that this larger output is differentiated from other partners who now also are able to compete in volume – the kind of volume we haven’t seen in human history.
The question then becomes, what does it mean to be Firm X or Firm Y? What’s different about how they negotiate warranties, or share purchase agreements? What qualities of their output gives the client the best possible result? In a situation where you’ve not codified what makes Firm X better than Firm Y, your legal operating system simply brings your work up to baseline and no further.
Historically, European law firms have been good at codification of judgement, investing in knowledge management teams and embedded practice support lawyers from the 1990s onwards. US firms, which tend to operate a different partner-profit-share model, have traditionally relied less on the central knowledge management function.
As a result, European firms may have a slight advantage in scaling that judgement, but US firms are catching up, with a massive amount of hiring into knowledge management teams. Whatever the location, partners should reflect on their desired output, and ensure that robust processes are put in place before amplification via AI ever occurs.
Supercharging knowledge management
Knowledge management teams are clearly incredibly important, but are often resource-constrained. Their responsibility to codify a law firm’s position, being a historically manual task. However, with AI also providing them with the means to turbocharge the amount of work they can do, they are increasingly able to keep pace with the data demands that generalist AI systems require.
Their work now defines why a customer should use one law firm over another, and without them, a firm is back at the median baseline of what a legal operating system can produce. There is of course the additional risk that without them, what’s amplified by legal OS tools is wrong if not properly codified before. The problem is clear with drafting – if a firm doesn’t codify its judgement and produce a gold-standard starting point and instead keep pulling down last year’s deal, replicating it (with or without AI), then it will simply amplify and entrench compromises that crept into the drafting during prior negotiations.
What’s required is investment in lawyer time to establish the best starting point for knowledge, before it gets anywhere near an AI. Legal data is no different to any other data: good data in means good data out. Firms must ensure their sources are correct and that the appropriate guard rails are in place, or they risk losing their authenticity.
Why does this matter now?
Beyond differentiation from competitors, the issue now in legal AI is that watermarks are coming. Without codification of proprietary firm knowledge, AI output risks being littered with watermarks, making it nigh on impossible to differentiate firm X’s product from firm Y’s (or even scarier for law firms, from alternative vendors entirely).
AI boosts the output a law firm can achieve like no tool that has come before it. But to keep ahead of the crowd, defining and protecting the firm’s position will always be the priority.
Join the conversation!