To get started with AI at work, pick one task you do by hand more than once, open the AI tool your employer approves, and give it the context a new coworker would need. Expect a draft that gets you about 80% of the way. The last 20%, your judgment, stays with you.
The first time I tried ChatGPT, I was not impressed. Someone on my team was using it for our communications, so I typed a request into the free version, got a mediocre answer, and decided I wasn’t going to pay for one more thing.
I spent 31 years at Ford, starting in a tool and die apprenticeship and retiring as a prototype operations manager. Trying new tools before anyone else was my job. I still almost walked away from AI after one try.
If one underwhelming try convinced you AI isn’t for you, start again with a better first task.
Why your first try with AI fell flat
Most people are figuring this out on their own. A Study.com survey reported by eWeek in January 2026 found that 89% of workers use AI on the job, but only about 33% got training from their employer. Nearly half, 46%, learned by trial and error.
Trial and error usually looks like one sentence typed into a chat box. You get something generic back, and you conclude the tool is overhyped.
A one-line request gives AI almost nothing to work with. Once I learned to give it context and go back and forth with it, the results changed.
Start with one task you already do by hand
The best first project is something you already do, more than once, that drains you. Ask yourself three questions:
- What do I do by hand every week?
- What do I put off because it’s tedious?
- What takes me 30 minutes when it should take five?
Here’s mine. After every free Learning Lab I ran, I had to download the Zoom recording, upload it to Vimeo, link it into the Notion page, write the follow-up email, figure out who attended, load it into Kit, and send a test. About 40 minutes, plus every distraction in between.
I opened Claude and said, let’s write down every step of this process first. Then we turned that process into a skill. The first time I ran it for real, 40 minutes of clicking was done just like that.
My mistake early on was creating systems for things I didn’t even do. Now I only build for what’s on my plate.
Most of us don’t have spare time to learn this. Microsoft’s 2025 Work Trend Index found that 80% of employees and leaders say they lack the time and energy to meet rising expectations. A repeat task is where AI gives you time back first.
AI should not become another thing on your plate. It should start taking things off it.
Practice at home, where mistakes are cheap
If work feels high stakes, start in your kitchen. I built a custom GPT I call my meal planner. It has my favorite recipes and a couple of cookbooks loaded in, and it knows my husband doesn’t eat as healthy as I do, so it adds a lighter option for me.
I ask it for next week’s dinners, approve four or five, ask for the grocery list, and my Instacart cart is loaded and waiting for me to approve it.
The more you use AI, the more you see problems differently. You hit something tedious at work and your first thought becomes, could AI help me with this?
Give AI the context a new coworker would need
Think about how you’d brief a smart new hire. You wouldn’t hand them one sentence and walk away. Give AI the same four things:
Who it should be. “You are a quality expert who has solved scrap problems and saved companies millions.”
What’s going on. “Our scrap was 1% and it’s gone up to 5%. We’ve already checked these three things.”
What you need. “Give me three things I can check and change today.”
One last line. “Ask me any clarifying questions before you answer.”
If you don’t like typing, use voice mode and talk it through. I ramble, and it still sorts out what I mean.
When you don’t understand the answer, say so. When I was in manufacturing engineering, I hesitated to take a profit and loss role because I didn’t know how to read a balance sheet. Today you can paste in a P&L and say, explain this to me like I’m in fifth grade, then ask what questions leadership will likely ask you about it.
Check the work, because AI can sound sure and be wrong
AI will get you about 80% of the way there. The other 20% is your responsibility, and you’re accountable for what goes out under your name.
AI can make things up and say them with total confidence. So push back. Ask it to play devil’s advocate, poke holes in its own answer, or score its work from 1 to 100 and tell you what it would take to reach 100.
Stick to the tool your employer approves, especially if you handle customer, patient, or financial data. At home, if you pay for an account, look in the settings and turn off the option that lets your data train the model.
Use AI to grow your career, not just clear your inbox
Plenty of people use AI to rewrite an email. Fewer use it to get promoted.
We forget our own wins. If your AI tool connects to your email and files, ask it to pull out what you worked on this year and attach the numbers. That list becomes your review, your promotion case, and your next set of goals.
A chef I know took over planning that someone had been typing into an unformatted Word document with two fingers. He used Gemini to help him automate a spreadsheet, then built an app for it in Replit. When they offered him a promotion with a number that was too low, he used AI to build his case for more and then rehearsed the conversation with it in voice mode.
You can do the same for an interview, a negotiation, or a hard conversation with someone on your team.
Don’t wait to be invited. The Women in the Workplace 2025 report from McKinsey and LeanIn.Org found that only 21% of entry-level women are encouraged by their manager to use AI, compared with 33% of men. Employees who are encouraged are over 50% more likely to use it. If nobody is encouraging you, consider this your encouragement.
If you manage people, go first
Managers set the tone. Use the tool yourself, talk about how you use it at home, and talk about where it got things wrong. That tells your team it’s safe to try.
I’ve seen managers pile on more work because they assume AI will absorb it. People end up buried. Ask instead what the team will do with the time AI frees up: more relationships, more mentoring, more of the work that needs a human.
Try this. Bring a real problem to a team meeting and brainstorm together how AI could help. Everyone’s judgment is in the room, and nobody learns alone.
If you’re weighing a new hire, read Before You Make Another Hire, Run the Job Through an AI Lens first.
Where to start this week
You don’t need a strategy to begin. You need one hour and one task.
Check what you’re allowed to use. Find out which AI tool your workplace approves. Many companies use Microsoft 365 Copilot. At home, ChatGPT, Claude, and Gemini all work.
Write down one repeat task. Pick the most tedious thing on your list and walk AI through every step of how you do it today.
Ask for a draft, then argue with it. Tell it what’s wrong, what you’d never say, and what’s missing. Keep going until it sounds like your work.
Keep a short list of what worked. Note the prompt, what you changed, and the time you saved. That list becomes your proof at review time.