AI in an SME starts with the leader, with an exercise to do before training the team

Four light bulbs in a row on a table, only the first one lit: the leader tries AI before the team

AI in an SME starts with the leader, with an exercise to do before training the team

In a small or medium-sized business (SME), the adoption of artificial intelligence (AI) starts with the person who runs it. As long as management does not use AI itself, AI remains a trial handed to the team. American author Geoff Woods made this the thesis of his book The AI-Driven Leader. He writes it in chapter 14: “You have to transform the leader before you can transform the organization”.

The first move for an SME leader with AI is to submit a real decision from the business to it. You ask the AI to question you before it proposes an answer. Three moves for the team follow: name a first adopter, start with volunteers, then say whether AI is a standard or a suggestion. Further down, you will find the exercise with a request to copy, then the Canadian figures that place your business.


Why AI adoption in an SME starts with the leader

AI adoption in an SME starts with the leader because the leader is the one who judges whether the tool is relevant. The leader also frees up the time to try it and sets what is expected of the team. In a business of 5 to 50 people, these three decisions can be made at a single table.

The article on the AI gap in small SMEs described the reasons for this gap from the company’s point of view. This one starts from a single person, the one who signs the decisions.

The author says the same thing from his first chapter. Getting educated about AI is not something you delegate to someone on the team: “Getting educated is your job” (chapter 1). He adds in chapter 2 that a leader who has seen what AI can do for them is then able to implement it in their business.

This condition is necessary, but it is not enough. The author does not promise that the team follows on its own once the leader is convinced. He then calls for work group by group, and a clear rule held by management. The second half of this article therefore covers those moves.

AI handed to the team while management watches

The scenario the author describes is one where AI is pushed down the organization while management works as before. Each team is told to explore. A few tools are tried, a few uses are noted. The way management thinks and decides does not change.

In chapter 14, the author calls this mismatch the leadership gap. Management asks its teams to change the way they work, but keeps deciding as before. The teams see it, and that is why AI stays a trial for them rather than a real change.

Canadian figures show a mismatch of the same kind in training. According to Statistics Canada, among businesses that used AI in the last 12 months, 32.0% reported AI-related training for their existing employees. For their existing executives, the proportion is 21.6% (Statistics Canada, second quarter of 2026). Even among businesses that have started, executives are less trained than employees.

In an SME, this mismatch can take a simple form. Someone on the team has opened a ChatGPT account to write emails or summarize documents. Management knows the tool exists without having tried it on its own decisions. This is called “shadow AI”, meaning employees using AI tools outside any framework set by the company.

The three obstacles that hold leaders back

If you have not done anything with AI yet, you are in the situation the author says he met most often. He says he interviewed more than 200 leaders to prepare his book. According to him, all of them believed AI was the future and that their company would adopt it, but fewer than 5% had done anything.

The useful part is the three obstacles he draws from them in chapter 2:

  • Time. Leaders were too busy with the day-to-day demands of their business.
  • Understanding. They did not know what AI is, or what it could do for them.
  • The path. No one had shown them a simple path from zero to a first useful result.

These obstacles describe a full calendar, a tool that is poorly known and the lack of a first exercise. The exercise below solves the third, since it gives that first path. It also answers the first, because it deals with a decision you have to make anyway and so adds no file to your calendar.

What Canadian figures say about AI in business

Canadian figures show that a minority of businesses use AI in their production, and that the reason most often given by the others is lack of relevance. Three public sources measure it.

Statistics Canada, second quarter of 2026. 19.2% of businesses reported using AI to produce goods or deliver services in the last 12 months. The proportion was 12.2% a year earlier and 6.1% two years earlier. The survey collected responses from 9,251 businesses or organizations, from April 1 to May 6, 2026 (Statistics Canada).

Statistics Canada, third quarter of 2026. 52.7% of businesses have no plans to use AI to produce goods or deliver services in the next 12 months. Among them, 79.1% feel AI is not relevant to them (The Daily, August 31, 2026).

Institut de la statistique du Québec. 12.7% of Quebec businesses used AI applications for production purposes, compared with 13.3% in Ontario. The period covered is the 12 months before the second quarter of 2025. The rate is 12.2% among businesses with 1 to 4 employees and 26.1% among those with 100 employees or more (ISQ, November 2025, in French).

One clarification is needed before quoting these figures. Statistics Canada measures AI used “to produce goods or deliver services”, and the ISQ measures AI used for production purposes. That is narrower than using an assistant like ChatGPT to write an email at the office. A business whose management writes its emails with ChatGPT can answer no to these surveys.

The most useful figure for a leader is therefore the 79.1%. In an SME, judging that a tool is not relevant is a management decision. That judgment can be made without having tried the tool on a real question from the business. The exercise below is a way to run that trial.

The exercise to do on a real decision in your business

The exercise consists of submitting to AI a decision you really have to make, with its context. You ask it to question you before it proposes anything. You judge the result yourself, because you know the decision better than the tool does.

The author calls this first spark the “lightbulb moment”: AI solves a real problem in front of the person who runs the business. In chapter 2, he advises looking for it in strategic thinking and decision-making, rather than in writing a better email.

His method fits in four letters, CRIT:

  • Context. You describe the situation in detail: your business, the decision, the options you see and what makes you hesitate.
  • Role. You name the expertise you want from AI, for example that of an advisor who has worked with SMEs in your industry.
  • Interview. The AI asks you questions, one at a time, before answering. It is the author’s favourite ingredient, because it draws out of your head what you would not have written.
  • Task. You state the expected result: options and their risks, a recommendation or a test.

Here is a request you can copy. Replace the brackets with your situation.

CONTEXT: I run a business of [number] people in [industry], in Quebec. I need to decide [the decision] by [date]. The options I see are [option A] and [option B]. What makes me hesitate: [your hesitation].

ROLE: Act as an advisor who has helped SME leaders with decisions of this kind.

INTERVIEW: Before proposing anything, ask me one question at a time, up to three questions.

TASK: Then give me the options I have not seen, the main risk of each, and the cheapest test that would tell me within a month if I am on the wrong track.

Four conditions protect the exercise. Open a temporary conversation, called an “incognito chat” in Claude, so that your account’s memory does not come into play. In ChatGPT, choose the unpersonalized option before the first message: it uses neither memory nor custom instructions. Also choose, in the model picker, the most powerful option your plan offers, because a business decision deserves more thinking than an email. Do not write any name or personal information. Answer the AI’s questions as you would answer an advisor, without preparing your answers in advance.

The decision belongs to you. The AI asks questions and proposes options, and you judge what holds up. The author describes the leader as the one who provides the context and decides, and AI as a thinking partner that questions and challenges.

What the exercise gave when I did it

I did the exercise myself on a real decision in my firm. I had to choose where to focus my sales efforts until the end of February. I am keeping my answers to myself, because they are about my clients. The AI’s questions, on the other hand, can be useful to you.

I used ChatGPT on November 14, 2026, in a temporary conversation, without naming any client. Between pasting the request and the moment I knew what I was deciding, I estimate I spent about 25 minutes.

The AI asked me three questions, one at a time, before proposing anything (my translation, the exchange was in French):

  1. “What have you managed to sell so far, concretely?”
  2. “Among your current mandates, which one would you like to reproduce ten times, and why?”
  3. “By the end of February, what matters most to you: quickly getting new paid mandates, or finding a client profile and an offer that you can then reproduce on a larger scale?”

I would not have thought of the third one. I was asking it a question about clients, and it first asked me to choose what I was aiming for.

What I decided: I am keeping a single type of client and a single offer, and I will measure the result by mid-December. That is the one-month test the request called for.

What matters for you in this result is the unexpected question. An AI that answers right away gives a general answer. An AI that questions you first works on your situation.

The three moves that follow the leader’s exercise

Once the exercise is done, three moves bring AI into the team: name a first adopter, start with volunteers, then say whether AI is a standard or a suggestion. They follow the adoption curve the author takes from the diffusion of innovations theory.

1. Name a first adopter

The first move is to choose one person on the team who will try AI in their daily work. You announce it to the team. In chapter 3, the book advises starting with the innovators and early adopters. You should encourage them to use AI every day and share with them what works.

The leader who has done the exercise has an advantage here. They can show their own conversation and the unexpected question that came out of it. To choose this person’s task, the article Where to start with AI in an SME offers four criteria and ten office uses.

2. Start with volunteers

The second move is to let AI spread from the people who want it, before aiming at the whole team. According to the curve the author uses, innovators make up 2.5% of a workforce and early adopters 13.5%. In a 25-person SME, these two groups together make 4 people (16% of 25, calculated from the book’s proportions).

These four people produce the first results the majority waits for before moving. For these results to convince, measure the time a task takes before and after. The article on the six-field grid gives a way to put a number on it.

3. Say whether AI is a standard or a suggestion

The third move belongs to management alone: saying whether the use of AI is a company standard or a suggestion. The author raises the question at the point where adoption has to extend to the last groups.

If it is a standard, management announces that this is the new way of working and that everyone is expected to evolve. It shares the results already obtained, offers support, then holds the standard. If it is a suggestion, everyone keeps the freedom to choose.

Both answers can be defended. What hurts is not deciding: the team does not know what is expected of it. According to Gallup, in May 2026, 25% of American employees say their organization has communicated a clear plan for integrating AI (Gallup).

What the leader’s exercise does not settle

The leader’s exercise is a starting point that leaves open the questions of data, accounts and usage rules. These questions come up as soon as the team works with real company data.

A temporary conversation, with no names or personal information, is enough for a management trial. A team’s daily work requires accounts in the company’s name, verified settings and written rules on what can be entrusted to the tool.

Nor does the exercise replace the leader’s judgment. An AI recommendation remains a hypothesis until the proposed test has given its result. That is why the request above ends with a one-month test.


What Groupe Ordinata does for SME leaders getting started with AI

Groupe Ordinata Inc. is an artificial intelligence consulting firm based in Montreal, specializing in process automation and AI agent integration for Quebec small and medium-sized businesses. We start from a real task, a number of minutes and a check, and the decision stays with the person who runs the business.

We bring more than twenty years of software engineering in industrial settings to this work, notably in real-time systems and machine vision.

When an SME moves from an assistant to an agent, accounts are opened in the company’s name and model training on its data is turned off. Depending on the tools selected, data may be processed outside Canada, notably in the United States. We tell you which tools and where your data is processed before we start. We also prepare with you, before implementation, the documents the law requires, such as internal policies and the privacy impact assessment (PIA). Since Law 25, this assessment is required, among other cases, before entrusting personal information to a provider located outside Quebec.


Move from the leader’s exercise to the team with an outside view

Our AI program begins with a $2,200 step before taxes, with no commitment to continue. We look at how your team works and draw up the first maps of your processes. Depending on what those maps show, the next steps may take the form of training on your own tasks, a template or an agent.

Prefer to start at no cost? Book a free 30-minute call, with no commitment. We discuss your tools and needs to identify where AI could save you time, what precautions to take and where to begin.

Book a free 30-minute call


Frequently asked questions about AI and the SME leader

Why does AI adoption in an SME start with the leader?

AI adoption in an SME starts with the leader because the leader judges whether the tool is relevant. The leader also frees up the time to try it and tells the team what is expected of it. According to Statistics Canada, 79.1% of businesses with no plans to use AI in the next 12 months feel it is not relevant (third quarter of 2026). In an SME, that judgment is a management decision.

What should an SME leader start with when it comes to AI?

An SME leader should start by submitting a real decision from the business to AI, with its context. The leader asks it to ask one question at a time before proposing anything. Use a temporary conversation, with no client names or personal information. You judge the result yourself, because you know the decision.

What is the CRIT method for writing a request to AI?

CRIT is the method of Geoff Woods, author of The AI-Driven Leader, for writing a request to AI. The four letters stand for context, role, interview and task. The context describes the situation in detail, the role gives the expected expertise. The interview asks the AI to ask questions before answering, and the task states the result you want.

How long does the leader’s exercise with AI take?

It depends on the decision, on the number of questions the AI asks and on the length of your answers. When I did it on a real decision in my firm, I estimate I spent about 25 minutes, from pasting the request to the decision. The preparation time is short: the decision and the context are already in your head.

How many Canadian businesses use AI in 2026?

According to Statistics Canada, 19.2% of businesses used AI to produce goods or deliver services in the 12 months before the second quarter of 2026. The proportion was 6.1% two years earlier. This measure is narrower than using an assistant like ChatGPT at the office. In Quebec, the ISQ measured 12.7% in the second quarter of 2025.

Should you train the team before the leader?

No: the leader goes first, then the team, starting with volunteers. According to Statistics Canada, among businesses that use AI, 32.0% trained their existing employees and 21.6% their existing executives (second quarter of 2026). A leader who has done the exercise can show their own conversation to the team.

Should AI use be mandatory in an SME?

It depends on what management decides, and management has to decide it. The author of The AI-Driven Leader puts the question this way: is AI a standard or a suggestion? If it is a standard, management announces it, offers support and holds it. If it is a suggestion, everyone keeps the choice.

References

  • Geoff Woods, The AI-Driven Leader: Harnessing AI to Make Faster, Smarter Decisions, companion PDF, page 1 (the 200 interviews); audio edition, chapters 1, 2, 3 and 14 (read in November 2026).
  • Statistics Canada, Analysis on artificial intelligence use by businesses in Canada, second quarter of 2026 (released June 11, 2026, read in November 2026): statcan.gc.ca
  • Statistics Canada, Canadian Survey on Business Conditions, third quarter 2026, The Daily (released August 31, 2026, read in November 2026): statcan.gc.ca
  • Institut de la statistique du Québec, Adoption et utilisation de l’intelligence artificielle par les entreprises au Québec en 2024 et en 2025 (released November 28, 2025, read in November 2026, in French): statistique.quebec.ca
  • Gallup, Global Indicator: Artificial Intelligence (May 2026 data, read in November 2026): gallup.com

Groupe Ordinata Inc., artificial intelligence consulting for small and medium-sized businesses, Montreal, Quebec. Workflow automation · AI agents · ERP/CRM integration · Data architecture.

Ordinata.ca | Contact us