Getting Started
How to Give AI One Good Example
Show beginners how one small example can improve AI output without turning the prompt into a giant template.
Tell the model who the output is for, what format you want, and one thing it must not get wrong.
Act as a patient work assistant. Help me with "How to Give AI One Good Example" for a beginner who needs a usable first draft.
Ask for a short version, one risk to check, and the next practical step. That keeps the result useful instead of vague.
Many beginners know they should “be more specific” with AI but do not know what that means in practice. One of the easiest ways to make a request clearer is to give the model one small example of the kind of answer you want. You do not need a long template or a complicated prompt method. A short example often gives the tool enough pattern to follow. If you are still choosing your first low-risk AI workflow, start with Getting Started With AI at Work.
Use an example when the format matters
Examples are most useful when you care about the shape of the answer, not just the topic. That includes cases like:
- short manager updates
- bullet summaries
- checklist formats
- plain-language rewrites
If the result needs to look a certain way, an example helps the model understand the target faster than abstract instructions alone.
Keep the example short and realistic
The best beginner example is usually tiny. You are not trying to do the whole task yourself before AI starts. You are only showing the pattern:
- one sample bullet
- one short paragraph
- one subject line and opening sentence
- one row of a table
Short examples work better because the model can copy the structure without getting distracted by extra content you do not really want repeated.
Separate the example from the real input
Beginners often paste an example and the real notes together without labeling them. That can confuse the model about what is source material and what is a pattern. A safer structure is:
- explain the task
- show the example
- paste the real material
Simple labels help:
- Example format:
- Real notes:
- Output:
That keeps the model from mixing your example into the final answer. If you need help getting the first rough request onto the page before you add an example, see How to Start an AI Prompt When You Feel Stuck.
Use the example to control tone and structure
An example is especially helpful when you want a certain tone or level of detail. For instance, if you want a calm internal update instead of a stiff formal message, one short example often teaches that style more clearly than saying “make it sound natural.”
The same is true for structure. If you want:
- one-sentence summary first
- bullets second
- next steps last
an example usually makes that much easier to reproduce. When you need the layout to stay stable from run to run, pair this with How to Ask AI for the Same Format Every Time.
Stop after one good example
Beginners sometimes overcorrect by pasting too many examples. That can make the prompt long, noisy, and harder to control. In most cases, one good example is enough to set the pattern. If the first answer is close, use a short follow-up to adjust it instead of adding more and more samples.
That is the main beginner rule: use one example to show the shape, then improve from there. You do not need a full library before AI becomes more useful.