09/03/2026
I have been thinking quite a bit lately about something that becomes increasingly obvious the more deeply I use AI in actual litigation: good legal research is not a linear process, and that may be one of the most important limitations of legal AI right now.
AI is extraordinarily useful for getting a lawyer oriented quickly. It can identify cases, synthesize lines of authority, summarize competing arguments, and help expose issues that deserve further investigation. I use those capabilities extensively. AI is amazing at capturing, comparing, and synthesizing clear propositions and major themes. But the subtleties are not always immediately captured.
Real litigation does not typically involve a clean question with a clean rule and an obvious answer. Most contested motions live somewhere in the gray areas — where several principles intersect, the cases are factually imperfect, and the outcome may depend upon precisely how a court understood the purpose of a rule or why it distinguished an earlier case.
That is where the lawyer still has to read the cases, understand the contours of the law, dive deeper into the courts’ analyses, and look for threads of analytical underpinnings that may help or hurt the client’s cause. When the lawyer reads the cases and engages with the dynamic movement of the court’s analysis as the court works through the issues, the research often changes direction.
We may discover that a case repeatedly cited for Proposition A was really concerned about Policy B. We may notice that the factual distinction the court emphasized looks remarkably like something in your own case. We may find an older authority buried in the citations that states the principle much more powerfully. We may realize that a later court technically distinguished a case rather than rejecting it — and that the distinction may actually strengthen our argument. Or we may simply recognize an analogy that no research platform was ever asked to find.
That last piece matters enormously. Good lawyers do more than locate rules and find cases with matching facts. They read between the lines. They notice what a court was really concerned about, what principle is doing the work beneath the stated rule, and where two lines of authority may connect even though no court has expressly connected them before.
Often the strongest argument is not the obvious one. It comes from recognizing that a court’s reasoning in a seemingly different context answers the same underlying problem, or from seeing an implication in a case that was never reduced to a neat proposition or headnote. That kind of legal reasoning requires more than retrieval and synthesis. It requires judgment, analogy, and the ability to connect dots that are not already connected for you.
Don’t get me wrong: I am not dismissing or diminishing the power of AI. I love my AI platforms and the ease with which they allow me to run down ideas, test arguments, and move quickly through complex issues. But I increasingly see the process as working with a team. AI removes burdens, accelerates the work, and makes incredibly meaningful — and sometimes very complex — analysis possible. But our minds still need to be the drivers: identifying what matters, shaping the questions, recognizing the nuances, and defining the contours of the argument. That is where the lawyer remains essential. Exciting times.
