When I started working with AI tools, I quickly realized something interesting. The results you get depend heavily on how you ask the question.
That may sound simple, but it is actually a shift in how we think about legal work. For years, we have focused on analyzing facts, applying law, and writing clearly. Those skills still matter. But now there is another layer. We need to know how to guide intelligent systems to produce useful outputs.
This is where prompting comes in. And in my view, prompting is not just a technical trick. It is becoming a core legal skill.
It can be tempting to assume that as large language models become more sophisticated, the need for careful prompting diminishes. In reality, the opposite is true. Most AI tools, whether directly or behind the scenes, operate on token-based systems, where both input and output contribute to cost and performance. Vague, keyword-style prompts (similar to a Google search) force the model to infer intent, often leading to longer, less efficient responses and increased token consumption. By contrast, well-structured prompts reduce usage costs while improving output quality. Prompting Is Structured Thinking
At its core, prompting is about giving clear, structured instructions.
That should feel familiar to lawyers. We already break down complex issues into components. We define scope. We clarify assumptions. We think carefully about language.
The difference is that now we are doing this not just for another human, but for a system that processes information differently.
When a prompt is vague, the output is usually vague. When a prompt is precise, the output improves.
For example, asking an AI tool to “summarize this case” will give you a general answer. Asking it to “summarize the key legal issues, the court’s reasoning, and the final holding in bullet points” produces something far more useful.
That level of clarity is something lawyers are already trained to do. We just need to apply it in a new context.
Thinking Like a Systems Designer
Prompting is not only about individual instructions. It is about designing how a process works from start to finish.
This is where I think lawyers need to start thinking more like systems designers.
A systems designer looks at the entire workflow. What is the input? What steps are required? Where does human judgment come in? What does the final output need to look like?
In legal practice, we often focus on the end product. A contract. A memo. A piece of advice. But with AI, the process becomes just as important as the outcome.
If you design a strong process, you can reuse it. You can refine it. You can scale it.
For example, instead of prompting once for a full contract, you might design a sequence. First you extract key facts. Then you identify risks. Then you generate clauses. Then you review and refine.
Each step builds on the last. That is systems thinking.
Breaking Down Legal Work into Components
One of the most valuable shifts I have seen is the ability to break legal work into smaller, more manageable pieces.
Traditionally, we approach tasks in a linear way. We gather information, analyze it, and produce a final document.
With AI, we can separate those steps more clearly. We can use tools to assist at each stage.
Prompting helps us define those stages.
For example, in a research task, we might start by asking for key issues. Then we ask for relevant cases. Then we ask for summaries. Then we compare those cases. Each prompt builds on the previous one. And using the AI model in this way, in a step by step approach, will lead to a better output response.
This approach not only improves quality, it also makes the process more transparent. You can see how you got from input to output.
That transparency is valuable for both lawyers and clients.
The Role of Iteration
Another important part of prompting is iteration.
You rarely get the perfect answer on the first try. And that is okay.
Good prompting is often about refining your instructions. You adjust the language. You add context. You narrow the focus.
This process is not very different from how we refine legal arguments. We test ideas. We revise drafts. We improve clarity.
The difference is speed. With AI, you can iterate quickly. You can test multiple approaches in minutes.
That creates an opportunity to explore ideas more deeply, but it also requires discipline. You need to know when to stop, when to trust the output, and when to apply your own judgment.
Where Judgment Still Matters
It is important to be clear about what prompting can and cannot do.
Prompting can guide a system to produce useful content. It can help organize information. It can surface patterns.
But it does not replace legal judgment.
Lawyers still need to evaluate the output. We need to check for accuracy. We need to consider context. We need to apply our understanding of the law and the client’s situation.
In other words, prompting enhances our work, but it does not remove responsibility.
This is something I emphasize often. The goal is not to delegate thinking to AI. The goal is to use AI to support better thinking.
Teaching Prompting in Legal Education
I believe prompting should be taught as part of legal education.
Not as a standalone technical course, but as an integrated skill.
Students should learn how to structure questions. They should practice breaking down problems. They should experiment with different approaches and see how outputs change.
This kind of training builds confidence. It also helps students understand the limits of the tools they are using.
In my own teaching, I try to create practical exercises. I ask students to work through real scenarios using AI tools. Then we discuss what worked and what did not.
These conversations are often more valuable than the outputs themselves.
A Shift in Mindset
Adopting prompting as a legal skill requires a shift in mindset.
We need to move away from thinking only about answers to thinking about how answers are generated.
We need to become more intentional about process. More comfortable with experimentation. More aware of how language shapes outcomes.
This is not always easy. Law has traditionally valued certainty and precision. AI introduces a level of variability that can feel uncomfortable.
But that variability also creates opportunity. It allows us to explore different approaches, to test ideas, and to improve how we work.
Building the Future Lawyer
When I think about the future lawyer, I see someone who combines strong legal reasoning with an ability to work effectively with technology.
They understand how to design workflows. They know how to guide AI tools. They are comfortable iterating and refining.
Most importantly, they remain grounded in the core values of the profession. Judgment, ethics, and responsibility do not change.
Prompting is just one piece of this evolution, but it is an important one.
If we can teach lawyers to think like systems designers, we are not just helping them use new tools. We are helping them build a more thoughtful, more efficient, and more adaptable way of practicing law.