What to Tell AI Before Asking for Help at Work

Beginners usually get weak AI answers for a simple reason: the tool is missing the context a coworker would normally ask for. If you only paste a rough instruction like “write this email” or “summarize this,” the model fills in the gaps with generic guesses. You do not need a complex prompt framework to fix that. You just need to tell the tool a few basics before you ask for help.

Start with the job, not the wording

Before you ask AI to write or summarize anything, tell it what job the answer needs to do. A useful answer depends on the task, not just the sentence.

The minimum starting context is usually:

• what you are trying to do
• who the output is for
• what kind of result you want back

For example, “Draft a short internal update for my manager” is already much stronger than “Write an update.” It tells the tool the audience and the situation before the writing even starts.

Add the details only you know

AI is good at filling blank space with plausible language. That is exactly why you should give it the facts that matter most before it starts. In work tasks, those usually include:

• the goal of the message or summary
• the important facts or notes it must use
• anything it should avoid
• the tone or level of detail you want

This is where many beginners go wrong. They assume the model will infer the right priority, level of detail, or audience expectation. It often will not.

Ask for a clear format

The shape of the output matters almost as much as the content. If you need a short email, bullet list, table, checklist, or three-sentence summary, say so directly. Otherwise the model may give you a long paragraph when you really needed something easy to scan.

Simple format requests work well:

• three bullets
• short email draft
• checklist with next steps
• one-paragraph summary

That makes the answer easier to judge and easier to edit.

Use one short setup example

Many beginners understand this faster when they see the context pieces in one place. A simple work prompt can look like this:

I need a short internal email to my manager explaining why a client summary is late. Use these points: we are waiting on final numbers, the draft is already done, and I can send the final version tomorrow morning. Keep the tone calm and direct. Limit it to 5 sentences.

That example gives the model the task, audience, facts, tone, and format. It is not fancy, but it removes most of the guesswork that causes generic answers.

Use a repeatable context pattern

You do not need to reinvent the setup every time. A simple pattern is enough:

1. tell the tool the task
2. name the audience
3. add the facts or notes
4. say what format you want
5. mention anything to avoid

That one habit improves most beginner prompts because it gives the model the context it cannot safely guess on its own.