AI Prompts That Actually Save Time at Work

Most prompt guides are written by people who find prompting interesting. Most people using these tools at work do not. They have a quote to send, a job post to write, an invoice to chase, and forty minutes before the next meeting.

So here is the short version.

Four things that make almost any prompt better

Give it a role and an audience. "Write a follow-up email" produces something generic. "You are following up with a facilities manager who asked for a quote two weeks ago and has not replied. Friendly, three sentences, no pressure" produces something you can send.

Show it one example of your own work. One previous email you were happy with does more than any amount of describing your tone. This single habit closes most of the gap between output that sounds like you and output that sounds like a chatbot.

Say what you do not want. No bullet points. No exclamation marks. Do not call the reader "valued." Constraints do more work than instructions.

Ask it to interview you first. "Before you write it, ask me any questions you need answered." This one line prevents most rewrites, because the tool stops inventing details it does not have.

Where the time actually is

The hour is rarely in the writing. It is in the starting. Four places where that shows up:

• The first draft of anything repetitive — quotes, proposals, job posts, service descriptions
• Turning a mess into a structure — meeting notes into actions, a rambling voice memo into a plan
• The email you are avoiding — the awkward one; the tool has no feelings about it and will simply produce a first version
• Reformatting — the same information for a client, a board, and a one-page summary

Build a small library, not a big one

You do not need a hundred prompts. You need six that fit your actual week, saved somewhere you can reach in five seconds. A note on your desktop is fine. The people who get value from these tools are not the ones with the cleverest prompts — they are the ones who stopped writing the prompt from scratch every time.

Two rules that keep you out of trouble

Nothing confidential goes in unless the tool is approved. Client names, account numbers, anything under NDA. If your organisation has not told you which tools are approved, that is a question worth asking out loud — and if you are the person who should have answered it, that is worth knowing too.

You own the output. If it goes out under your name, it is yours. Read it before you send it. These tools are confidently wrong often enough that this matters more than any prompting technique.

If you want a ready-made library organised by job rather than by tool, The AI Prompt Playbook is built for that. If you are the person who has to set the rules rather than follow them, AI Policy Templates is the place to start.