AI staff in the books: how small Icelandic companies multiply output without hiring

Most small and medium Icelandic companies face the same squeeze. There is more to do than there are people to do it, but it does not pay to hire a full position for a task that takes two hours a day. The bookkeeping, the VAT filings, the collections and the follow-ups all land on one or two people who have no time for anything else.

That is exactly where AI agents are starting to change the game. Not a chatbot that answers questions, but an automated worker that operates inside your systems: it reads invoices, posts them, tracks filings and flags when something is off.

What an AI agent does in operations

Think of it as a staff member who never takes a coffee break and never forgets a due date. Connected to Payday, DK or Regla, it can:

• Read incoming invoices, pull out the amount, VAT and date, and post them to the right account.
• Match receivables against payments, track who has paid and send reminders to those who have not.
• Watch the VAT period and warn you before the filing deadline, not the day after.
• Scan your sales data and point out customers ready for a repeat order or an upgrade.

Payday already has a fair amount of automation built in: recurring invoices, automatic reminders, payroll runs that calculate charges and file on their own. An AI agent sits on top of this and takes over the manual work that is left, the kind nobody wants to do at five on a Friday.

The numbers say most have not started

The report Staða gervigreindar 2026, based on a Statistics Iceland survey of around 850 companies, lays it out:

It is the small companies that have the most to gain, because they have the fewest people to spread the load across. Icelandic media, including Vísir, have covered how a large share of jobs will change with AI. The point is not that people get fired, but that the same team can produce far more if the machine handles the groundwork.

Where the human has to be

Here is the big thing too many skip. An AI agent should not run unsupervised in your finances. It is fast and never tires, but it can misread an invoice, post to the wrong account or misunderstand an unusual entry.

Human review has to be in these places:

• Before a VAT filing goes out. The machine prepares, the human approves.
• When an entry is unusual or large. Anything that breaks the pattern goes to manual review.
• At month-end and year-end close, where one error multiplies.
• When the agent itself is uncertain. A good system flags rather than guesses.

The rule is simple. Let the machine do ninety percent of the work and put the human on the ten percent where judgment and accountability matter. That is how you get the output without taking the risk.

Where to start

Start with one narrow task, not the whole ledger at once. Take invoice posting or collection reminders, run the agent alongside the old way for a few weeks and compare. Once you trust the result, add the next task.

This is exactly the kind of rollout, step by step into the systems the company already uses, that Vitvélar works through with Icelandic firms. Not an overnight revolution, but an agent that takes the dull work and hands the time back to people.

The practical takeaway: pick one recurring task in your bookkeeping this week, measure what it costs you in time, and try letting an AI agent take it over with a human review at the end. You will learn more in one month than from ten reports.