At a company of 10 to 25 employees, work is rarely neatly divided. The person who handles the administration also answers customer emails and draws up quotes. The person who does the purchasing also plans the logistics. Roles at this size are combinations of tasks that at a larger company would be spread across multiple positions.
That has consequences for the question of which work AI can take over. The role is not the unit against which that is assessed, but the task. One person can spend ten hours a week on work that AI could fully take over, and twenty hours on work that remains human work. If you look at the role as a whole, you don't see that difference. If you look at the tasks, you do.
At a larger company, a task is often already described: there is a process, a template, a colleague who also does it. At 10 to 25 employees, that knowledge often sits in the head of one person, built up through experience, not through a manual. That doesn't make a task unsuitable for AI, but it does make the first step different: before something can be transferred, it must first become clear what exactly happens, in what order, and based on what information.
That step is often skipped, with the result that automation gets stuck at a few isolated tricks — an email template, a spreadsheet — instead of at the underlying work. Work that has been made visible is easier to assess against the three categories that apply to every task: AI can take it over, AI can do it with human oversight that approves or rejects it, or it remains human work. At companies of this size, that breakdown is rarely made, simply because there was never time or reason for it.
At this company size, the barrier to automating something is rarely the technology. It lies in two other things. First: systems that don't talk to each other, so that data from bookkeeping, the CRM and email exist separately and someone has to manually retype or compare them. Second: capacity. At 200 employees, there is often an IT department or a process manager to set up something new. At 15 employees, the owner does that alongside everything else, or it doesn't happen at all.
That also explains why the picture varies so much between companies of this size. A company where the work largely consists of fixed, repeatable actions — invoicing, planning, standard correspondence — has a different automation profile than a company where a lot of customization per customer or assignment is delivered. And a company where much of the work consists of direct customer contact has yet other boundaries, because oversight and judgment weigh more heavily there than with internal administration.
At companies of 10 to 25 employees that already work with AI, it usually isn't about one large takeover of an entire role, but about pieces of work disappearing from the task list: part of the email handling, part of the reporting, part of drawing up quotes based on earlier quotes. What remains is often the part that requires judgment, negotiation, or specific customer knowledge.
That shift happens faster at one company than at another, and that difference rarely lies in the sector. It lies in how well the work itself is mapped out: which tasks there are, how much time they take, and what information they need. Companies where that is clear can answer that question per task. Companies where that isn't clear can only make an estimate at the role level, and that is a cruder and less reliable measure.
At a company of this size, every hour weighs more heavily than at a large organization, simply because there are fewer hours to divide. That is why we calculate not in roles or headcount, but in hours and FTE capacity — here we explain why that unit of calculation is more accurate than counting in people. An outcome expressed in hours shows how much capacity is involved in a task, without making any statement about who currently performs that task or should perform it. What an employer does with that outcome, moreover, falls under their own statutory requirements regarding personnel; we make no statement about that.
One last point that is often overlooked at companies of this size: not all data used in the work may simply be entered into an AI system. Before a task is assessed for transferability, it is good to know which data your company may not simply use, because that can directly affect the outcome of the assessment.
To know which part of the work in your company can be taken over by AI today, the work itself must first be mapped out: which tasks there are, how much time they take, and what they require in terms of data and judgment. The free quickscan offers an initial direction: twelve questions, no account needed, with an indication of which part of the hours in your profile could be taken over by AI today. The full work scan, which breaks down and assesses the work task by task across eight axes, is still under construction. What you can do today is use the quickscan to get a first picture of where in your company the freed-up hours are likely to be found.
Vraag maar. Ik ken de kennisbank van deze site; wat ik niet weet, zeg ik erbij.
Answers come from this site’s knowledge base. Not tailored advice, and not a scan of your company.