I introduced AI into developers' work and trained them to use it for concrete tasks. That taught me where value comes from: choose a problem, test a method, check the output and help people adopt the new workflow. I can bring that approach to the conversation between technology and business teams.
I use Codex, Claude, ChatGPT and other tools from OpenAI myself. In our development work, AI has helped with analysis, task breakdown, implementation and review. We estimated that AI made development about 50 percent faster in many tasks, including review and checking. This is an experience-based estimate from our team. See how we used AI in software development.
Find a task worth changing
If I am to help a company introduce AI, I start with the people who do the work. What takes time today? Which part requires judgement, and which part is repetition? Which data and confidential information are involved? What would be a better outcome for the customer or the employee? Those questions come before the choice of model or product.
My experience from customers' software projects is that technology rarely solves an unclear problem. That applies to AI as well. A small trial with a clear task, human oversight and an agreement on what we measure can yield more insight than a broad launch with vague expectations. When a trial works, we can describe the way of working, teach it to others and follow up on mistakes as well as gains.
Leadership means making the learning shared
When I introduced AI to developers, it was not enough for one employee to get an impressive result. The team had to talk about when the method worked, how output should be assessed and who took responsibility for the code that ended up in the product. We already had a culture of time for learning and knowledge sharing every other Friday. It gave us room to try out methods and share both good discoveries and mistakes.
That experience applies more broadly: people need to understand why a new workflow helps them and have the opportunity to ask critical questions. I want to give specialists room to challenge a proposal, because that is how we get a solution that lasts. It is also how you avoid mistaking fast output for good quality.
From trial to a workable way of working
After a trial, I want to hear from the people who will use AI day to day. Where did it save time? What did they have to correct? Their answers help us decide whether to change how the whole team works.
Contact: kontakt@tomc.dk · +45 299 297 67