How to find where AI is worth using in your firm

Guy PavlovPublished Updated

A task repeats every week or month, follows a pattern, uses information you already hold and is quick to check
Try AI on that task, and compare the hours before and after.
No task in your firm passes all four
Leave AI alone and put the time into other work. Asked what limits their use of AI, 40.0% of businesses told Statistics Canada it isn't relevant to the goods they produce or services they deliver.

I took the figure from Statistics Canada's Canadian Survey on Business Conditions for the second quarter of 2026, released on 27 May 2026.

"AI isn't relevant to my business" is a fair answer for some firms. To settle it for yours, test the tasks your team repeats. Start with the task that takes the most hours in a normal month, and ask four questions about it:

  1. Does it come round every week or month? A task done twice a year leaves few hours to win back.
  2. Does it follow a pattern? The same kind of information goes in, and the same kind of result comes out.
  3. Is the information already in email, documents or a system you use? If someone has to chase it first, AI has nothing to work from.
  4. Can someone check the result faster than doing the task by hand? A person in your firm reads every result before it goes anywhere, so the check has to be the short part.

Try AI on a task that gets four yes answers. Otherwise, move to the next task on your list. If no task gets four, leave AI alone and put the time into other work.

How many businesses say AI isn't relevant?

Statistics Canada's Canadian Survey on Business Conditions shows how common that answer is among Canadian businesses, and what else they report about AI:

How many businesses say AI isn't relevant?
Share of businessesWhat they reportedSurvey
40.0%Asked what limits their use of AI, said AI isn't relevant to the goods they produce or services they deliverSecond quarter of 2026 (opens in a new tab)
41.0%Among businesses with 5 to 19 employees, said AI isn't relevant to the businessSecond quarter of 2026 analysis (opens in a new tab)
13.4%Asked what limits their use of AI, cited cybersecurity or privacy concernsSecond quarter of 2026 (opens in a new tab)
19.2%Used AI to produce goods or deliver services over the previous 12 months, up from 12.2% in the second quarter of 2025Second quarter of 2026 (opens in a new tab)
52.7%Had no plans to use AI to produce goods or deliver services over the next 12 monthsThird quarter of 2026 (opens in a new tab)
79.1%Among businesses with no plans to use AI, felt AI isn't relevant to the goods they produce or services they deliverThird quarter of 2026 (opens in a new tab)

The 40.0% and 79.1% answers are about the goods a business produces or the services it delivers. Run your test on the work around the job as well: quotes, invoices, scheduling, replies to customers and the monthly figures someone prepares for you.

What does the test look like in a small firm?

To see how the test plays out, picture two firms whose owners both say AI isn't relevant.

The first is an electrical contractor with 12 electricians and an office manager. Nobody on site needs AI to wire a panel. In the office, the manager turns each electrician's site notes into a written quote, about 15 a week at 20 minutes each. That's about five hours a week on a task that repeats, follows one template and draws on notes and a price list already in a shared folder. Checking a draft against them takes about three minutes. The task gets four yes answers, so the owner tries AI on quotes for a month and compares the hours.

The second is a furniture maker with three staff, who build about five commissions a year. Each quote follows a long talk with the customer, and no two look alike. The bookkeeping already runs through accounting software. No task gets four yes answers, so the owner leaves AI alone and puts the time into the next commission.

What do you do with a task that passes?

Once a task passes, write down who does it, how long it takes, what they look at and what "done" means. The time gives you a number to compare against later, and my note on telling whether an AI tool is actually helping covers how to measure the change.

Then decide the next step: buy a tool, build something, use ordinary automation, change the process or leave the task alone. If client details would go into a tool, read my six checks before an AI tool touches client data first.

How do you check the rest of your firm?

The four questions look at one task at a time. A firm's answer also depends on the tools it already pays for, its data rules and who owns the result. The free AI readiness self-check covers all of these, along with repeated work, in 12 yes-or-no questions, and names the area to work on first.

For an answer on your own work, start with an AI Adoption Review. I map up to three workflows in your firm and write a recommendation for each one, which can be to leave the task alone.

To see which area to work on first, take the free AI readiness self-check.

Bring one task to a call.

Pick the one that takes your team too much time, and we'll work out whether a review makes sense. 30 minutes, free, no preparation needed. I'll tell you on the call if there's no opportunity.

Book a call

Guy Pavlov The call is with me.