How do you roll out AI training to a whole company?

To roll out AI training to a whole company, start with a short AI usage policy and a small list of approved tools, run a pilot with one or two teams, then train everyone else by role in short weekly steps. Keep a required safety module for all staff, make the rest practical and tied to real tasks, and track who is using AI and what it saves so you can adjust.

That is the short answer. Below is how to do it step by step, based on what tends to work in mid-size Canadian companies with mostly non-technical staff.

Why do most AI training rollouts stall?

Most rollouts do not fail because people are against AI. They stall for a few simple reasons:

• No clear rules. People are not sure what they are allowed to paste into a tool, so they either avoid it or use it quietly on personal accounts.
• Training is too generic. A one-hour webinar on “what is AI” does not tell an accountant how to use it on a month-end report.
• It is a one-time event. People attend, feel inspired, then go back to their inbox and forget.
• Nobody measures anything. Without numbers, leaders lose interest and the budget disappears.

A good rollout fixes each of these on purpose.

What should you set up before training anyone?

Do three things before the first session.

1. Write a short AI usage policy. It should say which tools are approved, what data must never go into them, and who to ask. Two to four pages is enough to start. In Quebec, Law 25 means you should involve your person in charge of the protection of personal information, and run a privacy impact assessment for any new tool that handles personal information.
2. Pick your approved tools. Many companies already pay for Microsoft Copilot or Google Gemini through their office suite. Others add ChatGPT or Claude business accounts. Fewer tools is better at the start. Make sure company accounts are set up so data is not used to train the tool.
3. Name an owner. One person, often in HR, learning or operations, keeps the rollout moving. They work with IT and a business sponsor from leadership.

How do you run a pilot that actually teaches you something?

Choose one or two teams with a clear, repetitive workload. Customer service, marketing and finance are common picks because they write, summarize and analyze a lot.

• Keep the pilot to 10 to 30 people and two to three weeks.
• Pick three or four real tasks per team, such as drafting customer replies, summarizing meeting notes or cleaning up a spreadsheet.
• Measure a simple baseline before you start: how long those tasks take today.
• Ask for feedback every week. What helped? What was confusing? What felt risky?

The goal is not to prove AI is amazing. It is to find out what training people need, which prompts work for your business and which risks show up in practice.

How should training be structured for different roles?

Split training into two layers.

Layer 1: required basics for everyone

This is short, around 30 to 60 minutes in total, and it covers:

• What AI tools are good at and where they make mistakes.
• Your policy: approved tools, what never to share, and how to report a problem.
• How to write a clear prompt: give context, describe the task, set the format.
• How to check output before using it.

Every employee should complete this and sign the policy before getting access to approved tools.

Layer 2: role-based paths

This is where the real value is. A sales rep, an HR advisor and a controller use AI very differently. Give each role a short path of 4 to 6 lessons with exercises built on tasks they already do. For example:

• Sales: preparing for a call, writing follow-up emails, summarizing a long RFP.
• HR: drafting job posts, preparing interview questions, writing policy summaries in plain language.
• Finance: explaining variances, writing formulas, summarizing reports for leadership.
• Customer service: drafting replies, building a knowledge base, handling difficult messages calmly.

Short lessons of 5 to 10 minutes, spread over a few weeks, work better than a single long session.

What does a good rollout timeline look like?

For a company of a few hundred people, a typical plan looks like this:

1. Weeks 1 and 2: write the policy, approve tools, set up accounts, name owners.
2. Weeks 3 to 5: pilot with one or two teams, collect feedback, adjust lessons and prompts.
3. Weeks 6 to 10: roll out department by department. Everyone does the basics first, then their role path.
4. Ongoing: share a monthly tip or new prompt, review numbers, and update the policy at least once a year.

For bilingual teams, offer training in both English and French from day one. In Quebec, this is often expected, and people learn faster in their own language.

How do you keep people using AI after the training?

Habits matter more than the first session. A few things help:

• A shared prompt library. Give people tested prompts for their role so they do not start from a blank box.
• Champions in each team. One or two people who like the tools can answer quick questions and share wins.
• Manager check-ins. A two-minute “what did you try with AI this week?” in team meetings keeps it alive.
• Visible results. Share real examples: “Customer service cut reply drafting time by a third.”

How do you know if the rollout is working?

Track a few simple numbers: how many people finished the basics and signed the policy, how many completed their role path, how often people use approved tools, and how much time they report saving on key tasks. Combine that with a short survey on confidence. If usage is low in one team, talk to them. The cause is usually unclear rules or training that does not match their work.

Where does Team Tutor AI fit?

Team Tutor AI was built for exactly this kind of rollout in Canadian companies. It gives you a ready AI usage policy with signature tracking, role-based training paths in English and French, a prompt library by department, and simple reporting on completion and adoption. You can run your pilot and full rollout in one place, without building the content yourself.