GenAI Prompting for Smart Legal Professionals

It’s as easy as asking questions!

When working with modern generative AI tools, creating effective prompts doesn’t have to be complicated or intimidating. One of the most powerful aspects of these systems is that you simply ask questions in plain English. In seconds, AI can analyze your request, find relevant information across your documents, and provide answers with links to source materials.

If you can ask a question, you can use AI effectively. This fundamental shift—from complex Boolean searches to natural conversations—transforms how legal professionals interact with information.

Prompting is Easy

Unlike traditional approaches to document review that require mastery of complex search syntax, effective prompting relies on skills you already have:

• Natural language works: You don’t need special commands—just type your question as you would if you were asking a colleague.
• Specificity helps, but isn’t required: Even basic prompts return valuable insights; adding detail only improves results.
• It’s a conversation: You can refine or follow up on answers to dig deeper, without starting over.
• Forget complex syntax: Unlike Boolean search, prompts are full sentences or questions—no proximity operators or arcane rules.

Prompting is like having a conversation with a knowledgeable colleague. The more you communicate naturally, the better the results.

Five Simple Prompts to Get You Started

Here are five simple information goals that demonstrate how easy prompting can be:

1. For finding key people: “Who were the main people involved in approving the project? For each, tell me their title or role and how they were involved.”
2. For timeline creation: “When did discussions about environmental compliance first begin and how did they evolve? Create a timeline of events in table format with columns for date, person, event summary and document reference.”
3. For identifying problems: “What concerns were raised about the product before it launched? List each concern in a bold heading and then summarize the discussions pro or con about those concerns in bullets beneath your headings.”
4. For understanding relationships: “What was the relationship between Sarah Johnson and Acme Corporation? How and when did Ms. Johnson become involved, with whom did she work, what roles did she have at the company and when did she have those roles?”
5. For comparing statements: “How did the company’s public statements compare with internal discussions about the merger? List each statement in a bold heading and then tell us how the internal discussions compared or contradicted each statement.”

These simple, conversation-like prompts yield meaningful answers rather than just search results requiring hours of review. As you become more comfortable in your prompting, you can enhance these simple prompts using more advanced techniques. But remember: Starting with a simple question is always a reasonable way to start.

Building a Prompt

While your prompt can begin as a simple question, you’ll get better results if you add a little structure to your request. An effective prompt typically includes:

1. Provide Context: Briefly explain the matter and relevant context.
2. Use Natural Language: Ask questions and make requests in a conversation.
3. Add Key Details: Include names, dates, or specific terms of interest.
4. Be Specific in Your Request: Clearly state what information you’re seeking.
5. State the Desired Output: Specify how you want information presented.

The following examples demonstrate how small adjustments in your prompt construction can dramatically improve the quality and relevance of AI responses. Study these to help develop your ability to craft precise, context-rich prompts across different investigative scenarios.

Providing Context

Contextual information helps Alchemy better understand your needs and the significance of different information.

Less Effective: “What discussions occurred about product safety testing?”

More Effective: “We’re representing a client in a product liability case involving the MediTech insulin pump. The plaintiff claims inadequate safety testing led to device malfunctions. Review documents to identify all discussions about product safety testing for the insulin pump prior to FDA submission, particularly focusing on any concerns raised by the engineering team or instances where testing protocols were modified or expedited. We need to understand if testing standards were compromised due to launch timeline pressure.”

Use Natural Language, Not Keywords

A conversational approach that includes as many keywords as possible in your request is better than keywords alone.

Keyword Approach: “financial irregularities accounting practices Q3 2022 CFO email disclosure”

Natural Language Approach: “Find communications where our CFO discussed potential irregularities in the accounting practices during the third quarter of 2022, particularly any emails about whether these issues needed to be disclosed to auditors or investors.”

Be Specific

The more specific you can be in describing what you are looking for, the better.

More General: “Find information about the contract breach.”

More Specific: “Find information about the Atlas Construction contract breach that occurred in summer 2023. Look for communications between our project manager Tom Wilson and Atlas representatives, particularly any mentions of missed deadlines or quality issues with the foundation work. Include any discussions about potential financial penalties.”

State the Desired Output

Unlike keyword searches where brevity is valued, Alchemy benefits from detailed prompts. State what you want to learn from your documents. Vague prompts produce vague results.

Without Output Instructions: “Identify all the witnesses who have discussed the manufacturing defect and what they said about it.”

With Output Instructions: “Identify all the witnesses who have discussed the manufacturing defect. Present your findings in a table with these columns: Witness Name, Job Title, Date of Statement, Key Points About Defect, Document Reference. Sort the table by date with the earliest statements first.”

Structuring a Prompt

You can add structure to your prompt, particularly when you want to offer multi-part requests or ask multiple questions:

Be careful when you copy and paste prompts using automatic numbering. These don’t transfer well to the prompt box. You may need to use manual numbers instead or add them once you have pasted in your prompt.

Enhancing Your Prompts

Once you have a sense of how prompts work, it becomes a simple matter to enhance your prompt and improve results. Here are several examples where we enhance a simple question to make it more effective as a prompt.

Prompting for Different Information Needs

Different investigation types may require different prompting approaches. The following examples demonstrate different but effective prompts for common investigative needs. We include them both to offer different approaches to finding information and to continue providing examples showing how easy and logical prompts can be once you move from keyword searches to simply asking questions.

1. Identifying Admissions

“Did any company representative acknowledge knowing about the safety issues in the Model X device before the recall announcement on June 15, 2023? For each person who made such an acknowledgment, provide their name, title, role, the exact nature of what they acknowledged knowing, when they claim to have learned it, and from what source.”

2. Identifying Missing Information

“Based on standard industry practices for pharmaceutical clinical trials, what information would typically be documented but appears to be missing from this document collection? Focus particularly on patient consent procedures, adverse event reporting, and statistical analysis methods.”

3. Understanding Complex Relationships

“What was the relationship between Sarah Johnson and XYZ Corporation between 2019-2022? Analyze how and when Ms. Johnson became involved with XYZ, with whom she worked directly, what formal and informal roles she held, and how her influence evolved over time. Identify any potential conflicts of interest in her various roles.”

4. Documenting Awareness

“When did key personnel first become aware of the accounting irregularities in the European division, and how did their understanding evolve over time? Create a progressive timeline showing when each person learned new information, what specifically they learned, and how their characterization of the issue changed in subsequent communications.”

5. Comparing Internal vs. External Communications

“How did the company’s public statements about the merger compare with internal discussions among executives? List each public statement in a bold heading and then provide bullet points showing how internal discussions compared to or contradicted each statement. Focus particularly on revenue projections and expected synergies.”

6. Contract Analysis

“What obligations did our company have under the Johnson service agreement? For each obligation, identify the specific contractual provision that creates that obligation and summarize any documented discussions about compliance challenges or interpretation disagreements related to each requirement.”

Selecting the Right Prompting Approach

You may choose a different prompting style depending on the stage of your investigation:

A Progressive Investigation Framework

Effective investigators often use a sequence of prompts that progressively deepen their understanding. This example demonstrates how to build an investigation through interconnected prompts:

Initial Assessment: “Provide an overview of all documents relating to the Alpha Project budget overruns. Identify key personnel involved, timeline of significant events, and major areas of concern.”

Targeted Follow-Up: “Based on the initial findings, focus specifically on the communications between the Project Manager and Finance Director between March-May 2023. What explanations were given for the cost overruns, and how did the Finance Director respond to each explanation?”

Pattern Identification: “Analyze the explanations given for budget overruns across all projects in 2022-2023. Identify recurring patterns, common justifications, and how the accuracy of initial budget projections compares across different project teams.”

Synthesis and Implications: “Based on all previous analyses, summarize the systemic factors contributing to budget overruns, the effectiveness of existing controls, and potential areas of concern for regulatory compliance or financial reporting.”

The best approach is to view prompting as an iterative dialogue with your document collection, progressively refining your understanding through thoughtfully constructed questions.

Prompting: Start with a Question

Throughout this guide, we’ve seen how AI transforms the traditional discovery process from painstaking keyword searching to natural, intuitive investigation. Whether you begin with a simple question or craft a detailed, multi-faceted prompt, AI works with you to uncover meaningful insights from your documents. The difference between traditional discovery platforms and AI isn’t just technological—it’s a fundamental shift in how legal professionals interact with information.

As you develop your prompting skills, remember that expertise comes through practice and experimentation. Start with straightforward questions, observe the results, and gradually incorporate more sophisticated techniques as you become comfortable. Each investigation becomes an opportunity to refine your approach and discover new ways to leverage AI capabilities. The most successful users view prompting not as a technical skill to master but as an ongoing conversation with their document collection—one that progressively reveals deeper insights and connections that might otherwise remain hidden.