How to Use AI at Work Without Making It Weird
- Start with your own writing, not anything client-facing. Low risk, immediate payoff.
- Check your company policy before you paste anything internal into a chatbot.
- The wins that get noticed are speed and clarity, not volume.
- Be open about using it. Hiding it is what turns a productivity story into a trust problem.
Most advice about AI at work is either breathless or paranoid. The reality is duller and more useful: there are about six tasks where it saves real time, and a couple where using it will get you in trouble.
Here is a five day version that gets you productive without a policy violation.
Day one: check the rules
Before anything else, find out what your employer allows. Many companies now have an approved tool and a written policy. Some ban public chatbots for anything work-related.
This is a ten minute conversation with IT or your manager and it prevents the one outcome that ruins the whole experiment. If there is no policy, assume nothing confidential goes in.
Day two: your own email and notes
Start where the risk is lowest. Take an email you already wrote and ask for a tighter version. Take your messy meeting notes and ask for a clean summary with action items.
You are editing your own words, so nothing confidential leaves that you did not choose to send, and you can judge the quality immediately because you know what you meant.
Day three: the thing you procrastinate on
Everyone has one recurring task they put off. The status report. The project update. The quarterly summary nobody reads.
Give the AI last month’s version plus this month’s facts and ask for a first draft. Procrastination is usually about starting, and this removes the start.
Day four: prepare instead of react
Before your next difficult meeting, describe the situation and ask what questions you should expect and what you might be missing.
This is the use that impresses people, and nobody can see you did it. You just show up having thought about it more than everyone else.
Day five: pick one to keep
Four experiments, keep one. The habit that survives is the one attached to something you already do weekly.
Trying to change six work habits at once has the same success rate it does in January.
How to talk about it
Say you use it. Quietly hiding it works right up until someone asks how you turned that around so fast, and then the story is about concealment rather than output.
“I had AI draft it and then rewrote the parts that mattered” is an answer that makes you look competent. It is also true, which helps.
The mistake almost everyone makes in week one
People start with the highest-stakes thing they do. The client proposal, the board summary, the presentation to the executive team.
It is the wrong place to start, because you cannot tell whether the output is good when you are anxious about it, and one bad experience there kills the habit for six months.
Start with the boring internal stuff nobody will see. Build judgment about what it does well on low stakes work, then take it to the things that matter.
Want the full foundation?
AI for Everyday Work is one of four books in the Core Four bundle: email without the dread, meetings and notes handled, writing anything faster, and the busywork killers. The bundle also covers getting started, research and decisions, and prompting properly.