I am fascinated by how many business problems turn out to be solvable with technology.
I challenge what everyone has stopped questioning, and I solve it with data-driven analysis and cutting-edge technology. Today that is AI. Three times the answer turned out to be a commercial model rather than a piece of engineering, and three times that became a company I built and sold.
I have sat on both sides of the table since: as the chief executive reporting to a board, and as a director on two. Increasingly a board has to form a view on AI without anyone in the room having done it before, and that is the kind of problem I want to be in the room for. What pulls me is the problem, not the size of the company or the title on the seat.
Frank MogensenCopenhagen
Most technology problems turn out to be business problems.
That is the habit I bring to a board: working out which problem the company actually has, settled before anyone agrees what to spend on it. It is a director’s question rather than a manager’s, and it is usually cheaper to ask than to skip.
- Founded
- Sold
- Capital raised
- Board seat
- 1999 Alpha IT Group founded
- 2007 Sold to Mondo
- 2013 Voopii founded
- 2014 Sold to Gigsky
- 2016 Cloudeon founded
- 2018 TDC takes 40%
- 2021 Sold to Devoteam
- 2023 CBS board exam
- 2025 Two board seats
Alpha IT Group
IT cost more every year, was configured by hand, and could not be budgeted. We automated the platform and sold what nobody was selling yet: infrastructure and operations at a fixed rate per employee per month.
Unpredictable project cost → a fixed price per head, a decade before that was normalSold to Mondo A/S in 2007 for DKK 30 million. Computerworld, June 2007Voopii
Roaming was priced so badly that companies switched it off when their people travelled — a capability bought and then forbidden. We assembled a service across several operators and sold local rates worldwide to enterprises and airlines.
A capability switched off because of price → local rates worldwideCloudeon
Three to five years from an idea to production, by which time the idea was outdated and the person who had it had left. Not a backlog — an operating model with a cycle longer than the shelf life of its own ideas. I was chief executive from the start and handed it over a year before the sale, to build the Lithuanian organisation myself.
Three to five years to production → days or weeksTDC took a 40% stake two years in. DLA Piper, August 2018 · Sold to Devoteam in 2021 at ~100 people. Computerworld, September 2021The same move three times. Each time the innovation was the commercial model, not the technology. The technology only made a different model possible.
Three quarters of large boards are deciding about AI without anyone who has done it.
These are not my numbers. Each one links to the study it came from, because a figure you cannot check does not belong on a page like this.
of listed companies above $1bn have a board with three or more directors who have actually worked with AI. The other 74% are deciding without one.
percentage points of return on equity above their industry average, for the companies whose boards have it. Boards without it average 3.8 points below.
of organisations can attribute any EBIT impact at all to AI — while 80% of the people inside them say it has improved their own productivity. The gap is organisational, not technical.
more likely to aim AI at product innovation and customer growth rather than only at efficiency — among the companies that actually outgrew their peers on revenue, margin and return on capital. Spending more on AI does not correlate with winning. Where you point it does.
of generative-AI projects are abandoned after the proof of concept — on data quality, risk controls, cost and unclear business value. None of those are model problems.
Challenge, and knowing where it stops
I question what has stopped being questioned. The change belongs to management; my part is being sure they can make it, and saying so when I doubt it.
What a board needs from management
I ran three companies with a board above me. I know what to ask for, and what a report looks like when it is comfortable rather than true.
Adopting AI, as a business decision
What it can be trusted with, what stays under human decision, and who answers when it is wrong. Three years on it, a chairmanship, two investments.
Risk, and technology risk with a number on it
Setting what the company will accept, and holding it there. Cyber and delivery risk included — where boards most often accept a figure they cannot interrogate.
Transactions, capital and succession
Three sales answered from the seller’s side, investment raised from TDC, and a chief executive handover I made myself. Most directors have only read about the last one.
Duties and reporting
Danish company-law duties, liability, financial oversight and sustainability reporting — examined at CBS, not assumed. Finance, cyber, people and operations come from running the companies.
Founder and chief executive of three companies, each with its own board and chairman. The part most technology candidates do not have.
Board member and investor at Workofo. Chairman and investor at Agencii, 2025–26. Both ended when I stepped back for personal reasons.
CBS Executive board exam, a six-month examined programme, 2023–24. Information Technology, Technical University of Denmark.
No competing board commitments, and able to commit for a full term. Nordio AI has its own chief executive; I am not in its daily operations. Past investments are disclosed before a committee meets, not after.
Banking, insurance, medical, retail, large organisations — as customer and as operator. Size matters less to me than whether the problem is the interesting one.
If you are putting a board together, or filling a seat.
Twenty-six years of being the person who had to deliver the plan is why I know when one is sound, and when it is optimism with a timeline attached. What the seat buys is oversight, not a second pair of hands on the wheel.
Write a few lines about what it concerns. You do not need to have finished thinking, and you will not get a form letter back.
Request a meetingThirty minutes, in your calendar