How to get employees to use AIYour team has AI. Here's how they'll use it.
Plenty of teams that pay for AI barely use it. That's rarely about the people. Here's why AI tools sit unused, what gets a team using them on real work and a 30-day plan to roll it out.
In short
To get employees to use AI, stop asking them to find uses for a new tool. Pick one annoying job they do every week, connect the AI assistant to the apps that job lives in and write down what's allowed. Try it with two or three people, train everyone on their own work, name one person to ask and show the team the time it gives back. Most teams can get there in about 30 days.
The gap between having AI and using it is wide. In May 2026, 47% of US employees told Gallup that their organization had integrated AI tools, but only 30% used AI a few times a week or more. Just 25% said their organization had communicated a clear plan for it. Buying the tool is the easy part.
Why employees don't use AI
It isn't connected to the apps they work in
If using AI means opening another tab, copying an email in, copying the answer out and pasting it into the CRM, most people will decide it's quicker to do it themselves. Often they're right. An assistant that can't see the order, the deal or the invoice saves less time than it costs.
The answers are generic
Out of the box, an assistant doesn't know your products, prices, policies or tone. So the first reply it drafts sounds like nobody in particular and gets a detail wrong. One bad answer is enough for many people to stop trying.
Nobody said what's allowed
Can I paste a customer's email in? A contract? Our sales numbers? When the answer isn't written down, careful people don't use it at all. In Gallup's February 2026 survey, 43% of non-users at organizations that offer AI named data privacy, security or compliance concerns as a barrier.
Fear of mistakes, or for their job
People worry about sending a customer something wrong with their name on it. Some worry about what AI means for their role: in early 2026, roughly 19% of US workers told Gallup their job was somewhat or very likely to be eliminated within five years by new technology, automation, robots or AI. Both fears go quiet when people can see that a person checks the work and that the point is to take chores off their plate.
No time to learn
Nobody has a spare afternoon to experiment. In the Federal Reserve Banks' Small Business Credit Survey, 37% of firms planning to use AI named the time to implement it or train employees as a challenge.
They can't see how it helps their work
"Use AI more" isn't a job. In the same February Gallup survey, 39% of non-users said they don't believe AI can help with the work they do, and 46% said they'd rather keep doing things the way they do them now. That's a reasonable reaction to a tool nobody has shown them on their own work.
What gets a team using AI
Start with one annoying weekly job
Ask the team which job they'd most like to stop doing by hand. Pull the Monday numbers, answer "where's my order?", prep for a call, chase a late invoice. One job, done well, beats a list of twenty ideas. Our list of everyday jobs AI can take on has examples.
Connect it to the apps they already use
Link the assistant to the inbox, help desk, CRM or books that job lives in, so it can look things up instead of waiting for someone to paste them in. Everyone signs in as themselves and sees only what their role allows. See the apps your AI assistant can work in.
Write down the rules
One page is enough: what the assistant can be used for, what never goes in, which tasks always need a person's OK and who approves them. Our AI usage policy template is a free starting point, and human in the loop explains how to decide what needs approval. Clear rules help: in Gallup's research, employees who strongly agreed their organization had clear AI policies were more likely to be frequent users.
Start with two or three people
Pick the people who do the job every day. Include one enthusiast and one skeptic; the skeptic finds the real problems. Give them a week on real work and fix what they find before anyone else sees it.
Train on real work, not on AI
Skip the generic prompt-writing course. Spend an hour with the team showing the assistant doing their own jobs in their own apps, then let each person try it on something from their inbox that day. Leave each role with a one-page cheat sheet of the requests they'll use most. If you'd like that done for you, see AI training for your team.
Name a person to ask
Every team needs one go-to person for "is this allowed?" and "why did it say that?". It doesn't need to be a technical person. It needs to be someone who knows the work and answers quickly.
Managers go first
People follow what their manager does, not what the memo says. Gallup found that employees who strongly agree their manager actively supports their team's use of AI are 1.7 times as likely to use it a few times a week or more. Use it in front of the team, for your own prep and reports.
A 30-day rollout plan
The plan suits a team of five or fifty. Stretch it if you need to, but don't skip steps.
| When | What happens | Who | Done when |
|---|---|---|---|
| Days 1–3 | Pick one weekly job. Time how long it takes today. | Manager, with the team | One job chosen, with a "before" time |
| Days 4–7 | Choose the assistant, set up business accounts, connect the apps the job needs. | Manager and whoever manages your apps | Everyone has their own login; the assistant can see the right information |
| Days 8–10 | Write the one-page rules: what it's for, what never goes in, what needs an OK and from whom. | Manager | Rules shared with the pilot group |
| Days 11–17 | Pilot: two or three people use it on real work every day and log what goes wrong. | Pilot group | A list of fixes, and a rough "after" time |
| Days 18–21 | Fix what the pilot found. Write a cheat sheet for each role. | Manager, pilot group | The pilot group would recommend it to the rest of the team |
| Days 22–24 | One-hour live training for the whole team, on their own work. Name the go-to person. | Whole team | Everyone has used it on one real task |
| Days 25–30 | Everyday use. Check in at the end of each week: what worked, what didn't, what to add. | Whole team, go-to person | Most of the team uses it for the job without being reminded |
At day 30, pick the next job and run the same steps, faster. Our page on building an AI plan shows how that adds up, one win at a time.
Show the time it gives back
The thing that turns a skeptic is seeing a colleague get an hour back. Make that visible:
- Time the job before you start and again after a few weeks. Use your own numbers, not a vendor's.
- Share the difference at a team meeting, along with what people did with the time.
- Count how many people use it each week, not how many have a login.
- Keep a short list of what it got wrong and what you changed. It shows the team their feedback matters.
If the numbers don't move, that's useful too. It usually means the assistant is missing information, the rules are unclear or the job wasn't a good fit. Fix that, or pick a different job.
Common questions
Why don't employees use AI at work?
Usually because it isn't connected to the apps they work in, its answers are generic, nobody has said what's allowed or they have no time to learn it. Some worry about mistakes or their job. Each of these has a fix, and none of them is "try harder".
Should we make AI use mandatory?
Making it mandatory before it's useful tends to produce box-ticking. Make it useful for one job first, train people on that job and let the results do the persuading. Then make it the normal way that job gets done.
How do I know if AI adoption is working?
Look at weekly use of the job you chose, the time it takes now against before and what the team says in check-ins. Logins tell you nothing. A history of requests, if your setup keeps one, tells you a lot.
What if someone makes a mistake with AI?
Treat it as feedback on the setup, not the person. Check whether a rule was missing or unclear, or whether the task should need a person's OK. That's why customer-facing work should be drafted by AI and sent by a person.
Do we need an AI policy before we start?
A short one, yes. One page that says what the assistant is for, what never goes in and what needs approval is enough to start. Our AI usage policy template is a free place to start.
Sources
- Gallup, Organizational AI Adoption Jumps Six Points (July 20, 2026): survey of 22,573 employed US adults, May 6–20, 2026.
- Gallup, AI in the workplace indicator: clear plan (25%) and manager support (1.7 times), as of May 2026.
- Gallup, AI in the Workplace: What Separates Adopters and Holdouts (April 12, 2026): survey of 23,717 employed US adults, February 4–19, 2026.
- Gallup, Using AI More Does Not Reassure Workers, Managers Do (September 9, 2026): job-loss worry as of the first quarter of 2026.
- Federal Reserve Banks, Small Business Credit Survey: 2026 Report on Employer Firms (March 2026): survey of small employer firms, September 3 – November 14, 2025.