Thomas Anglero

What happens now that the budget has been spent, the returns are being questioned, and someone has to decide what continues and what does not. A keynote for executive teams and leadership conferences.

You are being asked what your AI spend has produced, and you do not have a clear answer. Neither does anyone else, and nobody wants to be the first to say so.

This is not a failure of your organisation, or of yourself. For two years experimenting was the right thing to do. What has changed is that the permission to keep experimenting is disappearing. Cost and the lack of results decided it. Roughly half of senior leaders scaled back their AI agent deployments this year because what they were spending overran what they were getting back.

There are three major AI decisions in front of you and every one of them has ramifications. Which initiative gets commercialised properly, with a named owner and a line on the P&L. Which initiative gets stopped in public. And what you keep funding without a return, deliberately, with a future date at which that stops being acceptable.

I have been on both sides of this conversation. I approved budgets like these as Nordic CTO at Cognizant and Nordic Director of Innovation at IBM, and I now pay these costs myself in my company with an engineering team of AI agents rather than people.

This talk will leave you able to diagnose your own situation, sort your initiatives, and defend each one to your leader or your board with conviction and data.

A moat is only a moat from the inside. Your competitor sees a stream. For executive teams and boards, on who wins and who disappears in an AI economy.

Your business model, your moat, its advantages are real, and that is the problem. A moat you can see is a moat you stop checking. From the other bank it is a stream, and it is drying up. Today, AI is what dries it up faster. It lowers the cost of reproducing what you do, it removes the friction you have been charging for, and it gives a competitor the capability to disrupt what took you years to build.

I have worked inside two incumbent companies that thought their business model was perfect, and it was not. At Ericsson I argued in meetings that when the mobile connection got fast enough, the phone would stop being a phone. The leaders were not hostile to the idea. They simply did not think the mobile software mattered, because we owned the networks, we owned the handsets, and we owned everything in between. At IBM I founded the Watson AI Lab for Cancer at the Oslo Cancer Cluster, one of the first AI cancer laboratories on a European cancer campus. Building it was not the hard part. The hard part was that once it existed, there was nobody inside IBM I could take it to. No single owner of Watson, no one body that could carry it across the company. People who said they were in charge were not. I was told later that separate teams had been funded to compete with each other, and they were told not to share anything with each other.

Neither company was beaten by AI or a better product first. Both stopped being frightened first, and were beaten afterwards. Nobody had to attack. The moat had already dried up.

This is the part you can act on. Your competitor using AI is not the leading indicator. Your leading indicator is your own culture and your own processes, and it is visible long before anything appears on the horizon. Nobody in the leadership team is afraid. Disagreement has become expensive. The question of what would destroy you is no longer asked aloud by anyone holding a budget. Hubris, kingdoms, silos, and a flaw nobody is rewarded for naming.

I will also give you the opposite. A chief executive once asked me at my final interview whether I would go to war against his own organisation and find what was wrong with it. I said I would if he stood beside me. He did, and the work from that period is still running today. The day he left, it was dismantled. That is not a comfortable ending and I will not give you one, because this has to be owned at the top or it does not survive. The same is true of AI.

In AI there is a rule: bad data in, bad results out. An organisation works the same way. What you have built inside is what comes out, and AI is now amplifying it. You leave able to decide what gets fixed inside first, who must say the thing nobody is saying, and who you put in charge of the next round of AI. That is decided in your own building, and it decides everything after it.

Managing the change from AI experimentation to real adoption, and reshaping the way work is done across the organisation. A keynote for leadership teams whose first AI attempt did not produce the results they wanted.

Your organisation already has AI. What has probably not changed is how the work actually gets done. That is the challenge, and everything else is magnified by it.

The technology was always the easy part, and it is the part everyone was excited about. The hard part is asking your colleagues, and yourself, to work differently, when everyone is busy, when the last change programme did not go well, and when nobody can yet say what good looks like. Your people are not resisting AI. Most of them are waiting to be told what to do differently.

I have done this from both sides. I led AI and transformation across the Nordics as CTO at Cognizant and Nordic Director of Innovation at IBM, and I led innovation inside the Norwegian Tax Administration. Today I run a company with an engineering team of AI agents rather than people, so I am not describing this from a distance.

This talk is about where to start when AI has already been implemented. Who has to own it, and why it cannot sit with the technology function alone. What and who to change first, and what to leave alone for now. How to ask more of people who are already tired. And what it costs, honestly.

You leave with a plan of action, the order to do things in, the reasons behind it, and words you can use with your own people on Monday.