AI Adoption: How to Make It Stick After the Training

October 12, 2026

This is the video behind the post. It follows Owen, a made-up roofing owner, through all 3 steps, so you can watch the whole thing work before you try it with your own team.

AI adoption takes 3 things after the training is over. Each person knows which job they hand to AI. The wins are where everyone can see them. And one person owns keeping it going.

Training can fade unless those 3 things are true. This post is how to set up all 3 with your own team, in order.

You trained your team, and nothing changed

You ran the sessions. Everyone nodded along.

A month later, your office is typing every email by hand again.

The sessions are over, and so is the AI.

You’d think the job is done once the training is delivered. It isn’t. Delivered is not adopted.

So here’s the fix. Name the job. Show the wins. Give it an owner.

AI adoption, step 1: name the job

This step tells each person which job they hand to AI from now on.

A tool that belongs to nobody stays in the truck.

Write down each person’s name, with 1 job beside it. Ask each person, “What ate your time this week?” and pick the job from their answer. If your team already wrote down 3 ideas each for using AI, pick from those.

Say Owen runs a small roofing company with 6 people. This is a made-up example. His office manager picks the email that chases an unpaid invoice. His estimator picks the scope write-up after each site visit.

The job is handed over. The checking isn’t. AI drafts and people sign.

The mistake is telling your team to “use AI more.” The right move is 1 named job per person.

To go further: one job that crosses 2 people

Pick 1 job that two people share. In Owen’s company, the estimator writes the scope and the office builds the quote.

Set that job up in 1 Project in Claude, so both work from it. Claude is the AI chat we use as our thought partner. A Project is a space in Claude that keeps your instructions for 1 job. (Claude + Airtable shows one on a quote.)

Before you fill it, decide what goes in. The job, not a customer. No customer’s name, number or address. No real prices. No passwords.

Start with 4 lines. Say who does the job, and who it’s for. Then say what good looks like, and 1 thing it must never do.

If your Claude plan lets you share a Project, share that 1 with both people. Whether you can share depends on your Claude plan. If you can’t, both people paste the same 4 lines into their own Project.

Then test it on a made-up job. Both people ask for the same quote. Read both answers. If both follow your 4 lines, that job is set up. If 1 doesn’t, fix the 4 lines and test again.

For real quotes, your own rules for what goes into AI decide what gets cleaned first.

Keep that list of jobs. You’ll add to it in step 3. (Step 1 in the video)

AI adoption, step 2: show the wins

This step puts each win where the whole team can see it.

It’s like a before and after photo of a finished job. A win nobody sees can’t be copied by the next person.

Start 1 wins thread now, or use one your team already has. A wins thread is 1 group chat, wherever your team already talks.

Post the job and the time saved. Leave out customer names and prices. Everyone on the team can read it, so keep it to the win.

Owen’s estimator posts the scope AI drafted, and what he had to fix.

The mistake is keeping wins in your head. The right move is 1 thread the whole team can see.

If the thread goes quiet, post your own win and ask for 1 back.

Keep that thread. You’ll read from it in step 3. (Step 2 in the video)

AI adoption, step 3: give it an owner

This step makes 1 person answer for keeping this going.

Trucks get serviced on a schedule, not when someone remembers.

Put a 15-minute check-in on the calendar, once a month. Name 1 person who runs it. That can be you. Use the calendar you already have, and invite the whole team.

Here’s Owen’s agenda, and it’s only 3 things:

1. 1 new job handed to AI. Add it to your list.
2. 1 win shared. Read it out from the wins thread.
3. Reread your team’s rules page for AI, the one everyone agreed to.

No rules page yet? Write down what never goes into AI, and read that. Start with customer details, financials and passwords.

Small and steady beats 1 big push. The mistake is 1 more big training day. The right move is 15 minutes a month, with 1 owner.

If there’s no new job this month, say so, and keep the slot. (Step 3 in the video)

Why the order matters

Name the job. Show the wins. Give it an owner.

Skip step 1, and “use AI more” means nothing on Monday. Skip step 2, and each win stays in 1 person’s head. Skip step 3, and your office is typing by hand again.

Do this today

Write your list today, and book the first check-in.

You won’t know if it’s sticking today. You’ll see it at next month’s check-in. Today you set it up.

Once your team uses AI each day, the next move is building the system it runs on. If you’re deciding what to pay for and what to build yourself, build vs buy software runs 2 questions on that.

Post your team’s before and after in my Skool community. Leave out staff and customer names.

More at thevlway.com.

I’m in Your Corner. Now, Go Own It.