How to Use AI at Work: A Practical Guide

Most people's first real conversation with AI at work goes something like this: they type one long sentence into ChatGPT, get back something generic, and conclude "it doesn't really understand my job." Then they close the tab and go back to doing things the slow way.

The problem usually isn't the tool. It's that nobody explained the two or three habits that separate people who get real value from AI from people who bounce off it after one try.

Start with a job, not a question

AI tools are better treated like a new hire you're delegating to than a search engine you're querying. A search engine wants a short, precise question. A capable junior colleague wants context: what you're trying to accomplish, who it's for, what "good" looks like, and what you already know that they don't.

So instead of asking "write a project update," try something closer to: "Write a status update for my manager on the Q3 rollout. Audience is non-technical. We're 2 weeks behind because of a vendor delay, but the team caught up on testing. Keep it to 150 words, lead with the delay and the fix, end with next steps." That's not a longer prompt for the sake of it — it's the same information you'd give a person if you handed them the task.

Iterate instead of restarting

The biggest habit shift is this: don't throw away a bad first answer and rewrite your prompt from scratch. Correct it, the way you'd correct a draft. "Make this more direct," "cut the second paragraph," "this is too formal for our team" — each of these gets you closer in one step instead of three. AI tools keep the conversation in context, so correcting is almost always faster than restarting.

Use the tool that fits the tool you're already using

A lot of the friction in "using AI at work" disappears once you stop treating AI as a separate destination and start treating it as a feature inside tools you already use. Drafting in Word or Docs, summarizing a meeting in Otter or Zoom, triaging a backlog in Jira, building a first-pass dashboard in Looker Studio or Power BI — most of these already have AI built in, and it's usually faster than opening a separate chat window and re-explaining your context from scratch.

Know what not to hand over

AI is strong at first drafts, summaries, rephrasing, and finding patterns in text you give it. It's weak at anything requiring judgment about people, anything where being wrong is costly and hard to verify, and anything involving information it wasn't given (it will still answer confidently — that's the part to watch for). The habit worth building early is treating AI output as a draft you're accountable for, not a finished answer you forward along.

Where to actually start this week

Pick one recurring task that eats real time — a weekly report, a set of similar emails, meeting notes you always have to clean up — and try running it through an AI tool for two weeks before judging whether "AI at work" is useful for you. One well-chosen habit beats a dozen scattered experiments.

If you manage a team and want a structured way to get everyone doing this consistently instead of everyone figuring it out alone, that's exactly what our AI for Leaders & Managers course is built for.