“We don’t get hired for the floor” said the Kirkland & Ellis chairman last week after announcing the firm will spend $500 million to create its own AI platform. Legal AI raises the floor for all legal work, but clients expect Kirkland to deliver more than the minimum. This is the same for all the boutique firms I speak with.
Fortunately, you do not need millions of dollars to build a custom AI platform. You can customize general AI tools like Claude Teams or ChatGPT Business. Customizing these tools can be done in English, without the need computer programmers. You customize it like training a junior associate: explain a process, provide examples and give feedback. It will learn the patterns and adjust to your needs.
Before going further, it’s important to call out that you must use the enterprise plans, such as Claude Teams and ChatGPT Business. These plans have the security required for your client work and do not use your inputs to train the underlying models. The consumer versions are a different story and should not be used for client matters.
There is much to gain by using a general AI tool like this instead of an out-of-the-box legal-specific AI vendor.
The most important near-term benefit is flexibility. The payment terms are monthly and a fraction of the cost of the legal-specific vendors who often require an upfront annual contract. Using these tools allows you to run small, time-boxed trials to validate your return on investment.
You can also use these customized versions for finance, HR, business development and other business processes. Other vendor tools restrict coverage to legal processes.
The long-term benefits are equally significant.
The pricing trajectory for the general AI tools is going in the opposite direction as the pricing for legal-specific tools. OpenAI’s CEO has said costs are dropping 10x per year. Meanwhile, legal-specific AI vendors platforms are incentivized to increase prices over time. This process is well-documented by Research Contract Consultants. They increase renewal fees and move features into higher pricing tiers to force upgrades. Expect this to continue.
You also gain a strategic advantage by owning your customizations. How you customize your tools becomes your institutional knowledge. By contrast, legal-specific AI platforms want to replicate your customizations for all their customers. If you build your own, you retain the know-how and you cement your differentiation from other firms.
While the benefits are significant, the downside is that there is a higher burden on your team to make general AI tools good. There is real work to build out prompts and processes that get strong results. If you cannot dedicate several hours per month to build the tools that save you many more hours per month, then this is not an option you should pursue – legal-specific AI tools give you the fastest value. But if you elect to build your own, you can reach a higher ceiling.