Was there a job last year you wanted and didn’t bid, because your crew couldn’t have carried it? If there was, I’d run it through your head again, because the work didn’t change. Your AI capacity for small business did. My view is that with agent hours written beside people hours, a small firm can bid work that headcount used to rule out, and the larger firms are still too slow to stop you.
Most Canadian firms haven’t moved yet. Statistics Canada found 19.9% of businesses with 1 to 4 employees used AI in the year to spring 2026, and 40% of all businesses said AI isn’t relevant to them. That gap is the window. For an Alberta builder it is wider still, because construction sits at 9.2% nationally in the same survey.
The firms down the road got the same tools you did. What they also got was a longer approval chain. An owner can decide on a Tuesday and have the new setup running by Friday. A larger firm books a meeting about the meeting, forms a pilot group, and lands the change next quarter. Same software on both sides of the street. Very different calendars.
What changed about capacity?
Every job has two piles of work inside it. One pile only your people can carry: the judgment calls, the client conversations, the parts where your name takes the risk. The other pile has to happen around it: first drafts, document sorting, chasing paperwork, status updates, the first pass at research.
That second pile is where bids used to die, because it ate people hours you did not have. Agents can carry much of that pile now, under review, at a cost you count per finished piece. Agent hours are the hours in that second pile, done by the tool and checked by a person. Your buyers are already at ease with the shape. Many of them are prompting the first draft themselves before they ever call you.
Capacity used to be a headcount. Now it is a decision about which pile each hour belongs to.
Why can a small firm move first?
Because the decision sits with one person. When the second pile can be carried by agents, the question stops being if you have enough people. It becomes if you trust the setup enough to put your name on the finished work.
An owner can answer that in a day, test it on one job, and adjust by the next one. A larger firm answers through layers of approval, and each layer adds a week. That is the whole edge. The tools are equal now, and the calendars are not, and a small firm’s calendar is the shorter one.
A small firm’s edge this year is a fast yes from the person whose name is on the door.
Why will this window close?
Because the larger firms will finish reorganising. Their slowness is a phase, not a character trait. Committees do report in the end. Rollouts do land. When a firm with far more people can point agent hours at the same second pile, the old arithmetic comes back, and scale wins the way scale always has.
So treat the window as a window. Use it to win the work, deliver it well, and bank the client relationships while the bigger names are still in committee. A relationship won this year keeps paying after the capacity edge is gone. If I ran a small firm right now, I’d use this year to win the work the big firms are still meeting about.
How do you re-price the job you lost?
Pull the old bid, or rebuild it from memory if you never sent one. Go line by line. For each part of the job, ask who or what carries it now. Keep the judgment, the client contact, and the final review in people hours.
Move the drafting, sorting, chasing, and first-pass work into agent hours, and count the checking time inside those hours, because agent work nobody has checked is not finished work. That is the same habit as pricing the finished job: the number that matters is what a finished, signed-for piece costs, all in. Most owners who run this count find the job they lost fits inside the crew they already have, with the people hours left over landing on the parts clients actually pay for.
Do not bid the old way out of habit. The habit was written for a firm that no longer exists.
Which lost job should you re-price this week?
Pick one job you lost on capacity in the last year. Re-price it on paper, line by line, people hours beside agent hours. You are not bidding it again. You are learning what your firm can carry now, so the next job like it gets a yes instead of a polite pass.
Keep the sheet. The next time a bigger job lands on your desk, you will know within the hour if you can staff it, and knowing is most of the bid. Working out where agent hours honestly fit is part of the AI Strategy Roadmap.
More on AI capacity for small business
How does AI increase capacity for a small business?
Every job holds two piles of work. One only your people can carry: judgment, client conversations and risk. The other surrounds it: drafts, sorting, chasing paperwork and first-pass research. Agents can carry much of that second pile under review, which frees people hours for the first.
What are agent hours?
The hours of supporting work a job needs that an AI agent does and a person checks. Drafting, document sorting, status updates and first-pass research are typical. Count the checking time inside agent hours, because agent work nobody checked isn’t finished work.
Pull one lost bid this week and re-price it with agent hours beside people hours, checking time included. I break down numbers like that weekly in the AI Newsletter with ZAK.
5 Comments