Some thoughts on AI during my vacation

Spending some time away from work does not mean I don’t have to think about it. Not being part of the daily business means you can actually have some deeper thoughts and change your perspective on things. For example how work is organised at the moment, or what kind of role you want to play in the everyday business. How does the agency respond to current market trends (if at all) and what is our take on it? How do we see different perspectives as strengths and how may we create better alignment? What would be a good team setup for the future? What tools are we currently using and why? It felt good to reflect on questions like these although it makes me uneasy to realise the to do list gets longer and longer.

Why AI doesn’t work in your company (yet)

Everyone talks about it and every new digital product must at least claim it includes — AI, short for artificial intelligence. Companies like Open AI or Anthropic seem to push updates to their models almost daily now and I don’t know how many AI influencers claim they have the best “prompt” recipes to download. It is wild. Meanwhile an MIT report finds that 95% of AI pilot projects in companies fail to deliver measurable cost savings or profit increases. Now people only skimming headlines quickly conclude that the reason is that AI is failing or still not good enough to actually bring enough value for companies but nothing could be further from the truth. The main issue is not the tech itself but a “learning gap”. Many organisations and employees (still) don’t understand how to use AI effectively and therefore miss to design workflows that capture AI’s benefits while mitigating risks. For many the recipe was to just dump AI on existing processes rather than redesigning their processes with the power of AI. Startups often succeed more because they lack rigid legacy processes and this is a key insight. If you really want to reap the benefits of AI you have to zoom out, rethink, redesign and rebuild the way your company works and is organised at the moment. That people are rather opposed to change makes it even harder to introduce AI. Personally I think this is not only a struggle of big corporations but also one of the small companies and as a result of that we will see a lot of smaller companies going out of business over the course of the next three to five years, simply because they have not adapted to AI fast enough and cannot keep up with the competition at some point in the future.

When the prototype becomes the MVP

Over the course of my holidays I worked on some personal projects with a mixture of tools. I tried GPT-5 and rebuilt my website with it — that was as fast as it was amazing. I also continued to work on HeatMate, which is an app you can connect to your Viessmann heat pump to track its efficiency. HeatMate is a mixture of different AI tools plus custom development and design. I ended up using Claude Code directly from the terminal a lot and fed it GitHub issues. That worked really, really well. The more detailed your issues / prompts are, the better your output gets. Personally I am also quite interested in investing but found a lot of the investing tools and apps out there a way too complicated if you only want to find good stocks to put your money in. So I started building a little app and registered an account at financialmodelingprep.com to work with live stocks data as well as historical data. It works great and I will likely write more about it soon. I also used Claude Desktop to better organise folders on my Mac. Because it also allows you to connect to your Google Drive and your Google Mail you can also create pretty clever automations there too. Without going much more into detail here, I guess this gives a pretty good overview over how powerful AI is as a tool and as an addition to your business, whatever that actually is.

Will AI make digital agencies redundant?

When AI is so great and can accomplish so many things very fast, aren’t digital agencies irrelevant then very soon? That is a logical question when working a lot with AI (tools) and seeing first-hand what you can do with it. I think AI will definitely make those kind of agencies redundant that cannot or won’t keep up with the latest developments both in AI and technology in general. However those agencies that rethink their processes and services and on top of that gain proven expertise in these new technologies quickly will remain and shine. Tomorrow’s clients will demand more speed, more innovation, multidisciplinary thinking and faster execution. That is every agency’s challenge. To make your agency ready for AI I would ask the following questions first: What (process, habit, ritual, chore) can be automated? Which task takes a ton of time but is boring as hell? How would you do X if that person or tool would not be there? That last question should not lead to firing people but rather get your juices flowing to think of a new solution from the ground up, rather than just fixing a little thing here and there which doesn’t lead to a lot of impact most of the time. Apart from that traditional brainstorming sessions are dead and we would be better off with AI taking a seat at the table as it can organise our meetings better, get rid of all the fluff and straight to the point. This also leads to more time for real interpersonal connections rather than annoying meetings that nobody has time or energy for.