AI Explained Plainly home

Guide · · 4 min read

How to start using AI at work when nobody trained you

ProfessionalsLeadersService & Veterans

Your company started saying "use AI" at some point. Maybe there was an email. Maybe a coworker mentioned it in passing. But nobody sat you down and showed you what's allowed, which tool to open, or what it's actually good for.

That's normal, and it's fixable in a week. You don't need a course. You need the rules, one tool, and one boring task to try it on. (If "generative AI" is still a fuzzy term, here's what it actually is.)

Day 1: Find out what's allowed

Before you open anything, find out whether your company has an AI policy. Search the intranet or employee handbook for "AI," "generative AI," or "ChatGPT." Check what IT has posted.

If you can't find anything, ask. Here's a message you can copy and adjust:

Hi [name], quick question. I'd like to start using AI for a few
low-risk tasks, like first drafts of emails and cleaning up my notes.
Do we have an AI policy or an approved tool I should use? And is there
anything I definitely shouldn't put into one?

Asking first protects you. If there's a rule, you'll know it. If not, you've got a record that you asked.

Day 2: See if work already gave you a tool

A lot of people already have an approved AI tool through work and don't know it. A work account usually comes with better data terms than a personal one. Here's what the big three say, as of October 2026:

If your company uses Look for What the vendor says
Microsoft 365 Copilot Chat, signed in with your work account Microsoft says prompts and responses get enterprise data protection at no extra cost and aren't used to train its foundation models. A green shield at the top, next to New Chat, shows that protection is on.
Google Workspace Gemini in Gmail, Docs, or the Gemini app, signed in with your work account Google says Workspace doesn't use your prompts or content to train its models without your organization's permission.
ChatGPT Business or Enterprise A company workspace, set up by your employer OpenAI says it doesn't train on business data by default.

Two cautions. Your admin decides what's turned on, so you may not see every feature. And your company's policy still decides what you can put into a work tool.

If work gives you nothing and your policy allows personal tools, any of the big free assistants will do for now. ChatGPT vs. Claude vs. Gemini vs. Copilot can help you pick.

Day 2, continued: Learn the paste rule

Here's the short version: if you wouldn't email it to a stranger, don't paste it into a personal AI account. No customer details, health information, passwords, account numbers, or anything marked confidential.

If you're on a personal account, turn off chat training too. Issue #1 has the exact setting for ChatGPT, Gemini, Copilot, and Claude, plus what "training off" does and doesn't protect.

Day 3: Pick one boring task

Don't start with your hardest problem. Start with something you do every week, that's low-stakes and easy to check. Good first picks:

Not sure what to pick? Let the AI interview you:

I want to start using AI at work, but I'm not sure where it would help.
Ask me 5 short questions about my job and my week, one at a time.
Then suggest 3 low-risk tasks I could try with AI, ones where any
mistake would be easy for me to spot. Don't suggest anything that
needs confidential or customer data.

Day 4: Try it with a prompt that sets the rules

Most bad results come from one-line prompts. Tell it the job, the reader, and what it shouldn't do. This one works for most first tasks:

Help me with a work task. Use only the information I give you.

Task: [what you need, e.g. "a first draft of an email to a vendor"]
Who it's for: [the reader and how formal to be]
Facts: [your rough notes, with names swapped for roles]

Rules: Don't add facts, dates, or numbers I didn't give you.
If something's missing, ask me instead of guessing.
Keep it short and plain.

For a specific job done start to finish, see how to write a professional email with AI.

And if something at work confuses you, AI is a patient explainer:

Explain [term or concept] to me in plain English, like I'm smart
but new to this field. Give me one everyday example. Then tell me
what I should double-check with a coworker, because you might be wrong.

Day 5: Check everything, then decide

AI tools get things wrong while sounding sure of themselves. OpenAI's own help center says ChatGPT can invent facts, dates, quotes, and sources, and recommends treating it as a first draft rather than a final source. The same habit applies to every assistant.

So before anything leaves your hands:

  1. Check every name, number, and date against your notes.
  2. Read it out loud. Fix anything you wouldn't say yourself.
  3. Ask yourself whether you could explain every line if someone asked.

At the end of the week, ask yourself whether the task took less time once you'd checked it. Keep what helped and drop the rest.

Should you tell your manager?

I'd say yes, briefly. Something like "I've been using [approved tool] for first drafts of routine emails, and I check everything before it goes out." That's honest. It also shows you're careful, and it's the easiest way to find out whether your team wants you doing more of it.

If you serve

On duty, never paste classified information, CUI (Controlled Unclassified Information, the sensitive-but-unclassified category), personal data, or unit and operational details into commercial AI tools. For official unclassified work, use your service's approved platform; Issue #1 and Can I use ChatGPT in the military? have the details. Your command's policy wins.

AI Explained Plainly is not affiliated with, approved, or endorsed by the Department of Defense/Department of War, any military service, or the VA.

Start with Day 1 today

Search your intranet for "AI policy," or send the message above. Everything else in this plan gets easier once you know what's allowed.

Sources

All checked October 22, 2026.