Budget for AI like a tool, a skill and a habit.
When an organization decides to "budget for AI," the line that usually appears is a subscription: so many seats at so much a month. It is the easiest number to find and the easiest to approve. It is also the line that matters least to whether anything changes in how the work gets done.
The license is the part you can see.
A software subscription looks like the whole cost because it arrives as an invoice. Somebody signs it, finance codes it, and the tool shows up in everyone's browser by Friday. From the budget's point of view, AI has been bought.
What hasn't been bought is the time it takes a person to get good with the tool, or the attention someone has to give to keep it useful after the first month. Neither shows up on an invoice, so neither gets planned. They get taken instead, out of evenings, out of lunch breaks, out of the one person on the team who happens to be curious. In our work with teams, this is the most common reason a rollout stalls: the money was there, and the hours weren't.
So we suggest a budget with three lines. The tool is what you pay for. The skill is the time people spend learning to use it on real work. The habit is the ongoing attention that keeps it in use once the novelty wears off.
The skill line is paid in hours.
Training time is the line people most often leave at zero, on the assumption that staff will pick it up as they go. Some will. Most will open the tool twice, get a mediocre answer to a vague question, and go back to doing the work the old way.
Learning to use AI well on your own work takes practice with that work: drafting the intake letter, summarizing the board packet, checking the invoice batch, rewriting the job posting. That practice happens during working hours or it doesn't happen. Budgeting for it means putting a number of hours per person on the calendar and protecting them the way you would protect time for a required certification.
Say a team of twelve each gets two hours a week for six weeks. That is a real cost, and it should be written down as one, at the loaded hourly rate you would use for any other project. Writing it down does two useful things. It tells staff the time is sanctioned, so they stop feeling guilty for spending it. And it gives you a figure to compare with the license, which in most cases will be the smaller of the two.
If the training time isn't on the calendar, it is coming out of someone's evening, and it won't last.
The habit line is someone's attention.
The third cost is the one that keeps the first two from going to waste. After the training ends, someone has to notice which uses are working, collect the prompts and templates that people keep reusing, answer the questions that come up in week ten, and decide when a new use is worth trying. That is part of someone's job, whether you name it or not.
In a school district, it might be an assistant principal who spends an hour a week gathering what teachers have found and sharing the best of it. At a manufacturer, it might be the operations lead who reviews which drafts and reports the team now produces with AI and which ones slid back to the old process. Each week the work is small and easy to skip, so it needs a line in the budget like anything else.
Budget the habit line as a share of one named person's week, and plan to keep it for at least a year. When the habit is left as everyone's job, use tapers off, and the next budget cycle asks why nobody opens the thing you paid for.
How the three lines change over time.
The proportions shift as you go. In the first quarter, the skill line is heaviest, because people are learning and the hours are front-loaded. The habit line is steady from the start and stays roughly flat. The tool line tends to grow later, as specific teams discover they need a more capable plan or a second product for a particular job.
Organizations that put most of the year's money into licenses up front often find in month four that they have seats nobody opens and no time left to fix it. Planning the hours first, then buying tools in proportion to what people are actually learning to do, keeps the spend tied to use. It also gives you a natural checkpoint: when the skill hours taper off, look at what the team now does differently before adding anything new.
A town office or a small nonprofit may not have a separate technology budget at all. The same three lines still apply. They just come out of the operating plan as staff time and a modest subscription, and the conversation with the board or council gets easier when both are written down.
A three-line AI budget.
Template
Fill in each line for the next twelve months. Keep it to one page.
- Tool. Which products, how many seats, monthly cost, renewal date. Note who approved each one and what job it is meant to do.
- Skill. Hours per person per week, number of weeks, number of people, and the hourly rate you use for project time. Multiply it out and write the total next to the tool cost.
- Habit. The name of the person who keeps AI use going, the share of their week it takes (an hour, half a day), and the date you will review whether it is still needed.
Then add one line under the three: the date, about three months out, when you will compare what each line cost with what the team is now doing differently, and decide what to change for the next quarter.
