Secret Santa creates a strange little gift problem.
You may have a strict $25 limit, know the recipient only slightly, and still be expected to produce something better than a random checkout-lane object.
AI can help.
But only if the prompt protects what you do not know.
Otherwise:
coworker, 29, likes coffee
quietly becomes:
coffee obsessive who loves premium gadgets and cosy home decor.
That is not personalization.
That is fiction with wrapping paper.
Copy this Secret Santa prompt
I need a Secret Santa gift. Relationship/context: [coworker, friend group, family, club, etc.] Hard budget ceiling: [amount] Exchange date: [date] Facts I actually know: [details] Wishlist information, if any: [details] Important unknowns: [taste, size, home style, specialist preferences, etc.] Avoid: [workplace-inappropriate jokes, alcohol, fragrance, clutter, food restrictions, etc.] Give me six gift directions that stay inside the hard budget. Use only the facts I supplied. Do not infer personality, politics, health, relationship closeness or specialist taste. Keep the gift appropriate to the social setting and easy to receive. For each idea, tell me which known fact supports it and one thing I should verify.
The important parts are not complicated:
- the budget is hard
- the information is incomplete
- the group context changes what is appropriate
Tell AI where the gift will be opened
A good gift between close friends can be a terrible office Secret Santa gift.
The same joke changes when:
- your manager is in the room
- gifts are opened publicly
- you have known the recipient six weeks
- the workplace has clear policies
- the group has different ages and personal boundaries
So say so.
This is an office Secret Santa and gifts are opened in front of the team. Keep humour low-risk. Avoid anything about appearance, dating, alcohol, politics, health or personal hygiene.
That is more useful than:
funny Secret Santa gifts.
Use the wishlist if you have one
A wishlist is not cheating.
It is unusually good recipient data.
If the list contains several options:
Here are five wishlist items inside our $30 limit. Help me choose one based on the small amount I know about the recipient. Do not invent a more creative off-list replacement unless there is a clear problem with the list.
If the selected item leaves room:
The wishlist item costs $18 and the limit is $30. Suggest one optional add-on under $10 that connects naturally without making the gift more personal than the relationship supports.
Sometimes the correct answer is simply to buy the thing they requested.
A $25 budget is a ceiling, not a mood
Tell the model:
$25 is a hard ceiling including any add-on. If a good direction normally costs more, change the category instead of recommending a poor cheap version.
This matters with specialist products.
A bad $25 version of a $100 category is not a clever budget adaptation.
Change the category.
If you know almost nothing, stay easy to receive
When the information is thin, ask for directions that need less specialist knowledge:
I only know what I wrote above. Prioritise modest, useful, consumable or flexible gifts. Avoid specialist gear, clothing sizes, fragrance, decor and anything that assumes a private hobby or home setup.
The safer list may be less exciting.
That is fine.
Secret Santa does not require soulmate-level insight into Jordan from accounting.
Audit for awkwardness before buying
Once you have a shortlist:
Audit these ideas for Secret Santa risk. Check budget, social appropriateness, how much recipient knowledge each idea assumes, and whether the gift could feel too intimate, insulting or like extra work. Do not generate new ideas unless one clearly fails.
This is one of AI's better roles in a low-information situation.
Not more creativity.
A second pair of eyes.
Do not make the exchange harder than it is
If there is a clear wishlist item inside the limit, you may already be done.
If the group agreed on silly gifts, follow the theme.
If the tradition is intentionally small, do not use AI to escalate the stakes.
The point is to make Secret Santa easier.
Not to win it.