How long does an AI rollout actually take?

Plan for a season, not a week. The build I ran across my own team was 26 weeks end to end: a longer mastery track, a 5-week sprint where every person shipped one real working tool, and a research course. You can run the sprint alone in five weeks and get real tools out the door. The rest is depth you add over time. Anyone who tells you it is a one-day workshop is selling you a workshop, not a rollout.

What if half my team is skeptical or scared of AI?

Good. That is the honest starting point, and mandating over the top of it is how you lose people. The fix is to let each person aim AI at the task they personally hate, not the task you want gone. When the thing they build gives them back three hours a week, the fear turns into ownership. You do not argue people into AI. You let them prove it to themselves on their own worst task.

Do I have to buy new software or hire someone technical?

No. The whole rollout runs on the AI tools you already have and plain-English instructions. Nobody on my team wrote code. The skill you are building is knowing your own work well enough to hand off the right piece of it, and that is a skill your team already has.

Won't rolling out AI make people worried about their jobs?

It will if you aim it at their identity instead of their busywork. The whole point of the role-discovery interview is that each person decides what to automate, and they never pick the part of the job they are proud of. They pick the grind. Done right, a rollout does not shrink the team. It gives the team back the hours to do the work only they can do.

What is the difference between training my team and actually rolling AI out?

Training is watching. Rollout is shipping. This is the mistake that kills most programs: teams run a course, everyone nods, and nothing changes on Monday because nobody built anything. In my rollout every person shipped one working tool inside a five-week sprint. The shipping is the rollout. The training just gets them ready to ship.

How do I keep it from dying after the first burst of excitement?

Put it on a cadence and name an internal champion. A rollout with no rhythm is a one-and-done that fades by week six. Pick a standing time to share what people built, keep one person accountable for momentum, and re-run the sprint with the next task. Motion you repeat becomes culture. Motion you do once becomes a nice memory.

How to Roll AI Out Across Your Marketing Team (Without Losing Anyone)

Do not mandate AI and hope. Run Crawl, Ship, Scale: a sprint-first rollout where every person on your team ships one real working tool before you ever scale. Here is the plan I built across 26 weeks of training.

Most AI rollouts fail the same way. A leader mandates the tool, books one training, and waits for the team to change. Nothing changes, because nobody built anything they own.

Here is the fix, and it is a leadership move, not a tool tip. I call it Crawl, Ship, Scale: a sprint-first rollout where every single person on your team ships one real working tool before you scale anything. I ran this across my own team as part of a 26-week AI training program I built, and the ship step is the entire game. Training gets people ready. Shipping is what actually moves a team.

The core idea: you do not roll AI out by teaching it. You roll it out by having everyone build something.

Crawl, Ship, Scale

The whole framework is three moves, in order. Skip one and the rollout stalls.

Crawl. Get everyone literate and safe first. Enough hands-on fluency to use the tools without fear, plus the ground rules: a human signs everything that leaves the building, and if the AI is missing something it asks instead of guessing. Crawl is short. It is not the destination. It is the on-ramp.
Ship. Every person builds and ships one real working tool aimed at a task they personally hate. Not a demo, not a slide about AI, a thing they use on Monday. This is the step everyone wants to skip and the step that makes the rollout real. In my build this was a five-week sprint, and every person shipped.
Scale. Now you widen it. Share what people built, turn the best tools into team standards, and re-run the sprint with the next task. Scale is where a one-time push becomes a cadence, and a cadence is what becomes culture.

The reason most rollouts die is they try to jump from Crawl straight to Scale. They train, then mandate wider adoption, and skip the part where each person proves it to themselves by building. Ship is not the hard middle you rush through. Ship is the point.

Start with a role-discovery interview, not a mandate

Before anyone builds anything, each person runs a short guided interview on their own role. This is what turns a top-down order into personal ownership. You are not telling them what AI will take. You are helping them find the task they would most love to hand off.

Here is the interview. Give it to each person to run on themselves, in any AI tool.

You are helping me find the best task in my job to automate with AI.

Ask me these questions one at a time, and wait for my answer before the next:

1. Walk me through a normal week. What are the recurring tasks you do
   over and over?
2. Which of those do you dread? Which drain your energy for the least payoff?
3. Which tasks are mostly repeatable structure, where the thinking is light
   and the format stays the same?
4. Which tasks are the part of your job you are actually proud of, the
   judgment and relationships you never want to hand off?
5. Of the tasks I dread that are also repeatable, which one, if it took
   me 20 minutes instead of 3 hours, would change my whole week?

Then recommend the single best task for me to build an AI tool around,
and explain why in two sentences.

The magic is in question 4. It puts a fence around the work people are proud of, out loud, before you touch anything. People do not fear AI taking their busywork. They fear it taking their worth. Name the worth first, and the fear goes quiet.

The worked example: one person’s rollout

Here is what Crawl, Ship, Scale looks like filled in for one real person on a team. This is a composite, lightly cleaned up, with the task anonymized.

Phase	What they do	The output	Time it takes
Crawl	Learn the tool hands-on, learn the two ground rules	Comfortable running a chat tool, no fear	Week 1
Role interview	Run the guided interview on their own week	A ranked shortlist, one clear task to build	Week 1
Ship (build)	Build a saved tool for their most-dreaded repeatable task	A working first draft they can actually run	Weeks 2 to 3
Ship (harden)	Use it live, fix what breaks, add the “ask don’t guess” rule	A tool reliable enough to trust weekly	Weeks 4 to 5
Scale	Show the team, hand off the recipe, pick the next task	One tool becomes a team standard	Ongoing

Notice what the reader gets from this. By the end of week five, that person has not “learned about AI.” They have a working tool that gives back real hours every week, and they built it themselves. Multiply that by everyone on the team and you have not run a training. You have changed how the team works. If you want the deeper version of this task-sorting step, run it against The Task Audit first so you know each person is aiming at the right task.

The common mistakes (I hit most of these first)

Learn these the easy way.

Mandating without buy-in. A mandate over a skeptical team creates compliance theater. People open the tool because they were told to, use it for nothing real, and quietly wait for the mandate to fade. Buy-in comes from letting each person aim AI at their own worst task, not from a memo.
Training without shipping. This is the big one. Teams run a course, everyone nods, and Monday looks exactly like Friday because nobody built anything. If your rollout ends in a certificate instead of a working tool, you did not roll anything out. You held a class. Make shipping the deliverable.
Automating away people’s identity. Point AI at the part of the job someone is proud of and you will get resistance that looks irrational but is not. It is self-protection. The role interview exists to fence off the pride work first, so people automate the grind and keep the craft.
No internal champion. A rollout with no owner drifts. Name one person whose job is momentum: they keep the cadence, celebrate what got built, and unstick the people who are stuck. Without a champion, the rollout is a leadership hobby, and hobbies get dropped when the quarter gets busy.
One-and-done with no cadence. The first sprint creates a burst of excitement that fades by week six if nothing follows it. Put the rollout on a rhythm. Re-run the sprint with the next task, keep a standing time to share builds, and the motion becomes culture instead of a memory. This is the same reason an AI marketing workforce compounds: you never stop repeating the one move that worked.

What to say when you take this up the chain

You will likely need leadership to fund the time, and “we want to use more AI” gets a shrug. What gets a yes is a sprint with a shipped outcome and hours saved attached to it. Frame it as a five-week sprint where every person ships one working tool, then report the hours the team got back. That is a number leadership can approve. If you need the full script for that conversation, here is the pitch that works.

Do this this month

Pick your five-week window. Run Crawl in week one: get everyone literate and hand out the role-discovery interview. Spend weeks two through five on Ship: every person builds and hardens one real tool. Then hold one meeting to Scale it, where everyone shows what they built and you pick the next task.

You do not need the whole team bought in on day one. You need one honest sprint where people build something they actually use. Buy-in is not the thing you get before the rollout. It is the thing the rollout earns you, one shipped tool at a time.