I build bespoke AI employees — limited roles that are carried out on an ongoing basis, built around your routines and handed over to run in your organisation.

It is a different kind of order from an automation: an automation performs a step when it is called, an employee has a standing assignment.

What is unusual about this is that I run my own in live use, every day, and have done so long enough to know what breaks after three months.

Nyheter AI is an editorial team of AI agents that monitors 27 sources around the clock. Glänne & Söner is a one-person business that still manages publishing, coaching, talks and development — and it manages because I have built a team of AI employees that handles parts of the work. The product is also available as a service of its own, My AI team.

That does not mean it suits you. It means I can tell you what it costs to keep running, which is the part that almost never appears in the quote.

Automation or employee?

Automation performs a step

Something comes in, something happens, something goes out. It starts when it is called and is finished. The quote that is written, the email that is sorted, the report that is compiled. That is usually the right choice, and it is cheaper — see the page on automation.

The employee has a standing assignment

It monitors, follows up and keeps things in order without anyone having to ask. The difference shows when the task has no clear starting point: "keep track of when the contracts expire", "summarise what has happened in the industry this week", "remind us of anything that has been unanswered for more than two days".

It fits what falls between the chairs

Almost every organisation has work that everyone agrees should be done and nobody owns. It gets done in bursts, when someone feels guilty. That is the kind of task a role suits, because the problem is not that it is difficult but that it is never prioritised.

And it costs on an ongoing basis

Unlike an automation that lies idle between runs, an employee is working all the time, and that shows in the running cost. It should be in the quote from the start — otherwise it often turns up as a surprise a few months in.

If you first want to understand what an agent is, and where it goes wrong: What is an AI agent – and what does it do in a small business?

How a role is built

We describe the role as if we were recruiting. What it is to do, how often, what it must never do, who it reports to and how you can tell it has made a mistake. This sounds laborious, and it is the part that decides everything — an unclearly described role becomes an unclear employee, just as with people.

It is built against your systems and your exceptions. The basic case is seldom difficult. What takes time is that a business is full of "except when it concerns that customer".

A person at the last step. Before anything leaves the building or affects a customer. It removes most of the time without moving the responsibility, and it is the right trade-off in almost every smaller business.

You get the log and the keys. The assets are in your name, the documentation is yours, and the handover is included. The aim is for you to manage without me.

And a follow-up after a month. That is when you see whether the role actually does what you expected. About half the time it needs adjusting, and that is normal — it applies to new hires too.

When you should not build one

When the task has a clear start and a clear end. Then an automation is cheaper, simpler and less to maintain. Do not build a role where a single step is enough.

When nobody is to own it. An AI employee needs someone who notices if it stops working, just like everything else in operation. Without such a person it will fall silent on a Tuesday in February, and nobody will notice for three weeks.

When the role is already unclear for you today. If nobody can describe what is to be done and when, it cannot be built — and then that question is the one to solve first. AI on top of an unclear way of working makes the unclear thing faster, not better.

Roughly one in five of my first calls ends with me advising against it. That is cheaper for us both than finding that out after a project. The full range of services is on the main page about AI consulting services.

Who you are hiring

I have a Master's degree in psychology from Örebro University, work as a speaker and coach, and have published a non-fiction book. In parallel I build and run a number of digital services — among them My AI team, which builds bespoke AI employees for small and medium-sized businesses, and Nyheter AI, where an editorial team of AI agents monitors 27 sources around the clock.

Glänne & Söner is a one-person business. That it still manages publishing, coaching, talks and development is because I have built an AI team that handles parts of the work. It is the same kind of system I build for others — and because I run my own in live use, I know what breaks after three months.

The combination of psychology and technology is unusual for an AI consultant, and it is the whole point. What decides whether an AI project succeeds is seldom the model. It is whether the people who are going to use the system understand why, and feel they gain something from it.

More about me

Common questions about AI employees

What is an AI employee?

A limited role that is carried out on an ongoing basis by an AI system instead of by a person: it monitors something, compiles something, responds to something or keeps something in order — every day, without being reminded.

The difference from ordinary automation is that automation performs one step when it is called, whereas the employee has a standing assignment and works without anyone starting it.

What does it cost?

Builds are quoted according to scope. A limited role usually comes to a few half-days to build.

On top of that comes an ongoing operating cost for the services it uses, normally tens to a few hundred kronor a month for a smaller business. You receive a fixed price for the build itself after a first free meeting, before the work starts. Benchmarks are in the article on what an AI consultant costs.

Does it replace an employee?

No, and I do not sell it that way.

In almost every assignment I take, the alternative is not an employee but for the work not to be done at all — the monitoring nobody has time for, the follow-up that gets forgotten, the compilation that would have been good to have. An AI employee takes on what falls between the chairs, not what someone already does well.

What happens when it makes a mistake?

It will make mistakes, and a build that does not handle that is not finished.

Roles are therefore almost always built so that a person approves before anything leaves the building or affects a customer. You also receive a log of what it has done, so that it is possible to see what happened and when.

Which role is it that never gets done?

Describe it in a couple of sentences, and I will tell you whether it can be staffed this way and roughly what it would come to. Free of charge.

Book a first meeting