The tool is not the transformation.

The most common opening question we get is which AI tool a firm should buy. It is a reasonable question and it is almost always premature, for a reason that becomes obvious once stated: the tool is the last decision in the sequence, not the first.

Consider what a purchase decision requires you to know. Which named action are you automating? What is its monthly volume? How long does it currently take? What does a good output look like, specifically enough that you could tell a vendor's demo from your own reality? A firm that has done its documentation can answer all four in a sentence. A firm that has not will describe a desire, and will end up comparing products against each other rather than against a job.

That comparison is where evaluations go wrong. Legal AI products are not general-purpose in practice, however they are marketed. A tool that is genuinely excellent at contract review may be mediocre at litigation drafting and useless at research retrieval. Without a named action to test against, you are assessing demo polish, which correlates weakly with whether the thing will work on your matters.

There is a further reason we stay out of the recommendation business, and it is a principled one. The primitives are the firm's asset. The matter definition, the action library, the role chart, the evidence rule, the research stock - those belong to the practice and outlive any vendor. Tools are how those primitives run today. A studio that ties its methodology to a particular product has quietly converted the client's operating model into the vendor's asset, and the price of that gets paid at renewal.

So the honest answer to 'which tool should we buy' is usually: not yet, and here is what to do in the meantime. It is a less satisfying answer than a recommendation. It is also the one that produces a practice which can swap its tooling in a year without rebuilding anything, which is the outcome that actually matters.