How to write a prompt

Start with a clear objective

Before you type a single word, ask yourself what you actually want to accomplish. Are you trying to summarize a document? Extract specific data? Generate ideas? The clearer your objective, the better your prompt will be.

Consider the difference between these two prompts:

The second prompt tells the AI exactly what to produce, how to format it, and what to prioritize. This specificity leads to responses that match your actual needs. When you're asking questions effectively, clarity is your greatest asset.

Use specific, descriptive language

Vague prompts lead to vague results. When you want something specific, say so explicitly. Instead of asking the AI to "look at" a document, tell it exactly what to look for and what to do with that information.

For example, if you're working with a quarterly report and need key metrics, try this approach:

The specific prompt guides the AI toward exactly the information you need. Adobe Acrobat AI Assistant helps users interact with documents by understanding these detailed requests and delivering focused responses based on the actual content of your files.

Context matters

AI systems work best when they understand the broader picture. Providing context helps the AI frame its response appropriately and deliver more relevant information.

Consider how context changes the output:

The second prompt gives the AI crucial information about your audience and purpose, resulting in a summary tailored to your specific situation. When working with document prompts, context transforms generic responses into genuinely useful outputs.

Context can include information about your role, your audience, the purpose of your request, relevant background information, or constraints you're working within. The more relevant context you provide, the better the AI can tailor its response.

Output format specification

Telling AI how you want information presented is just as important as telling it what information you need. Different formats serve different purposes, and specifying your preferred format upfront saves time and produces more usable results.

For instance, if you need to document meeting notes, compare these approaches:

By specifying the exact structure you need, you get output that's ready to use immediately rather than requiring additional editing and reorganization.

Iteration and refinement

Effective prompts often evolve through refinement. Your first attempt might not produce perfect results, and that's completely normal. Treat prompt writing as an iterative process where each response helps you improve your next request.

Common mistakes to avoid include:

When working with a document summarizer, you might start with a general request and then follow up with more targeted questions based on the initial response. This back-and-forth approach often yields better results than trying to craft one perfect prompt from the start.

1. Be specific about your desired outcome

The more precisely you describe what you want, the closer the AI response will match your needs. Instead of asking for a general summary, specify exactly what aspects matter most and how detailed the response should be.

Compare these two approaches:

The specific version tells the AI exactly what to focus on, how many points to include, and how long each should be. This level of detail consistently produces more useful results.

2. Provide relevant context

Context shapes how AI interprets your request. Include information about your audience, purpose, and any constraints that affect what you need.

When using prompts to chat with your documents, adding context like "I'm a project manager reviewing this contract for timeline risks" helps the AI prioritize information relevant to your specific role and concerns.

Context can be as simple as stating who will use the information or as detailed as explaining the broader project the request supports. Either way, it helps the AI deliver more targeted responses.

3. Break complex requests into steps

When you need something complicated, resist the urge to cram everything into one massive prompt. Sequential prompts work better than overloaded single requests.

For example, if you're analyzing a contract for specific terms, try this approach:

Breaking the task into steps allows you to verify the AI's understanding at each stage and adjust your approach if needed.

4. Review and refine your prompts

Treat your prompts as iterative drafts rather than one-time instructions. If the first response doesn't quite hit the mark, analyze what was missing and adjust your prompt accordingly. You might ask for more detail on a specific point, request a different format, or provide additional context that helps narrow the focus.

Over time, you'll develop a sense for what works and build a personal library of effective prompts for your common tasks.