Good prompting is less about clever wording and more about supplying what the assistant cannot see: your context, your constraints and your material. A vague request produces a generic answer, so name the situation and what you need - "I have 10 minutes before this meeting, what are the three things in this report I will be asked about?" beats "summarize this." Attaching your own sources keeps answers grounded in real material rather than invention, and asking a follow-up when the first reply misses is normal, not a failure. If you want help getting started, reach out to aihelp@syr.edu.
Writing better prompts
Overview
The quality of what you get back tracks the quality of what you ask. Give the assistant real context and a real constraint rather than a bare instruction: "Summarize this" gets you a generic summary, while telling it your situation and what you actually need gets you something you can use. Bring your own material - attach the PDFs, documents, spreadsheets or slides you are working from - so answers are grounded in your sources instead of guesswork. Then iterate: ask again at the level you need until the answer lands.
Steps
Give real context and a real constraint. Say who you are, what you are trying to do and what a good answer looks like, instead of just "summarize this."
Bring your own material. Attach the PDFs, Word documents, spreadsheets, slides, images or code you want answers about, so the response is grounded in your sources.
For work you return to often, keep the background in place (for example, in a Project) so you are not re-explaining it every time.
Iterate. If the first answer misses, ask again - narrow the ask, correct it, or request the explanation at the level you actually need.
Official links
Still need help?
If this did not answer your question, submit a request and the ITS Data and AI team will follow up.
Contact ITS AI Support