Is your business ready for AI?

AI readiness means one specific task in your business is well enough understood, and its data tidy enough, that a machine could take part of it and someone could tell whether the result was right. It is a property of a particular task rather than of the company as a whole. A business can be entirely ready for automated invoice sorting and completely unready for anything touching customer records.

You need clean data for the specific task, not across the whole business. If you want AI to draft replies to common customer questions you need a decent record of past questions and good answers. Your inventory data being a mess is irrelevant to that task. Waiting for company-wide data quality before starting is the most common way businesses postpone AI indefinitely.

For off-the-shelf tools, no. The skill that matters is process knowledge: understanding what a good output looks like and being able to spot a wrong one quickly. Statistics Canada found 32.0 percent of AI-using businesses provided AI training to existing employees, and 44.4 percent made some staffing or training change, which suggests training the people you have is the common path rather than hiring specialists.

A good candidate happens repeatedly, follows roughly the same shape each time, has an output a person can check in under a minute, and currently consumes hours you can count. If any of those four is missing the task is harder than it looks. The check-in-under-a-minute test matters most, because a task whose output takes as long to verify as to do saves nothing.

Then fix that before buying anything. This is the most common blocker and the least technical one. When a process is split across three people who each know their part, nobody can describe the whole thing, so nobody can say what the automated version should do. Assigning one owner costs nothing and resolves more stalled AI projects than any tool selection.

Statistics Canada found 19.9 percent of businesses with 1 to 4 employees used AI in the 12 months before its Q2 2026 survey, against 27.8 percent of businesses with 100 or more employees. Very small firms adopt AI at rates much closer to large ones than their resources would predict, probably because the decision is faster and the process being changed is simpler.

Enough to run a real trial and to pay for the staff hours it consumes, not enough to fund a transformation programme. The software is usually the smaller number. Statistics Canada recorded cost as a barrier for 10.6 percent of businesses in Q2 2026, ranking behind cybersecurity and privacy concerns at 13.4 percent, which suggests price is not the main thing holding Canadian businesses back.

Then choose a first task that does not touch personal information, and there is almost always one. Internal scheduling, supplier document summaries, and drafting standard non-client correspondence are common examples. Starting on the privacy-free side of the business lets you learn how this works before you take on the harder compliance questions covered in the risk chapter.

The tools change monthly, but the underlying skill you are building does not: identifying a repetitive task, measuring it, changing a process, and checking output quality. That skill transfers to whatever tool exists next year. Waiting protects you from tool churn and costs you the learning, which is the more valuable of the two.

Below 9 out of 20 on the four-part scoring below means the constraint is not AI. Usually it is an undefined process, an absent owner, or a task too varied to be worth automating. Buying a tool at that point adds a subscription to an unsolved problem. Fix the named weak area first, then rescore, which typically takes weeks rather than quarters.

Rescore whenever you consider a new task, because readiness is task-specific rather than a permanent company state. Also rescore after any significant process or system change, since a new CRM or a reorganised team can turn a previously unready task into a straightforward one. There is no value in an annual company-wide readiness audit.

For a low-cost off-the-shelf tool on a task with no personal information, yes, and that is often the fastest way to learn. The assessment earns its time when the project involves connecting systems, spending real money, or handling personal information. In those cases the questions below surface the blocker before it costs you a quarter.