--- title: "We were early three times: what 95 guilders in 1992 taught me about AI in 2026" author: Marc Diks date: 2026-05-20 modified: 2026-06-24 category: AI & Strategy reading_time: 9 min url: https://www.marcdiks.nl/en/blog/early-three-times-what-95-guilders-taught-me canonical: https://www.marcdiks.nl/en/blog/early-three-times-what-95-guilders-taught-me language: en --- # We were early three times: what 95 guilders in 1992 taught me about AI in 2026 An [advertisement, a website before Google](/en/blog/ai-recent-data-search-engine-determines) existed, and the shift to online: three moments when we simply did it. And that is exactly what I now see missing. In 1992, a few months after completing my military service and a brief stint at Amev Levensverzekering, I joined my father's business. Diks Verzekeringen, a local insurance brokerage in IJsselstein that he had started in 1970 by selling glass and contents insurance door to door, as thousands did across the Netherlands back then. One day an inspector from Aegon was sitting at our table, who first looked at that Beetle on the driveway — my first car — and then at the stack of car policies on the desk, and he remarked that in the Netherlands hardly anyone was actively offering classic-car insurance. My idea was to place a three-line [advertisement](/en/blog/dark-patterns-ban-cancel-button-2026) in the classifieds of the Saturday edition of De Telegraaf. We are talking pre-internet, and anyone who wanted to sell a second-hand car back then placed a small ad there. My father was fiercely against it, because he thought 95 guilders was far too much for such a tiny notice. He let me try it anyway. That Saturday our phone rang off the hook with people who wanted more information or wanted to take out classic-car insurance straight away, and we had nothing ready: no brochure, no quote form, nothing, because we were completely unprepared for this. On Monday it was even crazier and we closed dozens of classic-car policies. We kept that advertisement running for ten years afterwards and sold hundreds of policies with it. It was not vision and not a master plan, it was simply doing and seeing what happens. I have since seen that same attitude work twice more, and that is precisely the attitude I now see missing in SME directors looking at AI. ## Early, round one: a 95-guilder classified ad What that advertisement did was actually nothing special. It was not technology and not innovation, it was simply finding a path no one had walked yet. Other intermediaries sold classic-car insurance too, but no one advertised it actively in the channel where classic-car owners happened to look: the classifieds. That match made the difference. Because we were the only ones present on Saturday, and because we actually answered the phone — the office was at our home — we got customers from across the country. My father had a point that 95 guilders was a lot of money for a three-line ad. The rational decision would have been to do some homework first: market research, a business case, deciding on a pilot budget. We did not do that. We took a gamble we could afford, and watched what happened. Over ten years Diks Verzekeringen grew from a local intermediary offering all kinds of insurance into a nationally operating automotive specialist. Not because we had a strategy deck, but because we happened to buy three lines of text and let them run. ## Early, round two: a website before Google existed In 1997 we put our first website online: diks.nl. That was well over a year before Larry Page and Sergey Brin formally [founded Google](https://about.google/company-info/our-history/) in September 1998, and long before most Dutch people even had an internet connection. The timing sounds visionary now, but here too the same law applied: it was not vision, it was simply doing. After the classifieds experience my father had learned something, and this time he was the one who saw that the internet could become big. He felt we had to be there, even though we had no idea what we were going to do with it. What we did quickly realise: if customers were looking for classic-car insurance, they had to be able to find us at the moment they were searching, not on a Saturday in the newspaper. A few years later visitors could calculate and take out classic-car insurance online with us. As far as I know we were the first in the Netherlands to offer that. The difference between 1992 and 1997 was not the market, but the channel. The channel was now always open, 24 hours a day, and reached the whole of the Netherlands without a single brochure having to be sent. People who did not even know we existed found us via a search query. That leverage was new. ## Early, round three: parting with the field force In 2008 my sister and I, who were then running Diks Verzekeringen together after taking it over from my father, made a decision that was considered odd in the industry at the time. We parted with our field advisers. In the insurance industry the field force was sacred back then. Advisers drove around to businesses and private individuals, drank coffee, went through policies, and built a relationship that was meant to justify the premium. Many fellow intermediaries found our switch unwise, to put it politely. Some warned that we would lose our customers to offices that still visited in person. In practice the opposite happened. We could now invest the time and money previously spent on a roving field force into better website functionality, into online marketing through focus, and into services that were more relevant to our target group. That target group was by then private individuals with a classic or ordinary car, who were sitting at home behind a computer and wanted clarity quickly, not a coffee visit. At some point we received so many car claims from customers with only third-party cover that we began offering the recovery of those claims from the liable party as a free service. From that grew 112schade.nl, with which we also helped non-customers recover claims free of charge. At its peak we processed more than 1,000 claims a month there. In 2018 my sister and I sold Diks Verzekeringen to Voogd & Voogd Verzekeringen, with whom we had worked closely since 2001. Voogd had bought the advisory arm of VSP and was looking for someone to run it, and we merged Diks and VSP-advies. With the merger of Voogd and the Heilbron group in 2021, [Alpina Group](https://www.alpina.nl/over-alpina/) was formed, where I grew from Diks Verzekeringen into cluster director Online. ## What I see now with AI That same three-part pattern I saw over twenty years at Diks Verzekeringen (there is a new development, you do not know whether it will work, you do it anyway) is what I now see again with SME directors looking at AI. Only in 2026 a lot of it sits on the wrong side. [Research by Sharp and Wolters Kluwer](https://www.wolterskluwer.com/en/news/dutch-smes-are-leading-the-way-in-europe-in-terms-of-ai-ambitions-and-cloud-infrastructure) shows that 74.4% of Dutch SMEs have integrated AI into their operations, putting us well above the European average of 63.9%. That sounds impressive. But when you look deeper, that percentage falls apart. [Research by Van Spaendonck Groep](https://www.mkbservicedesk.nl/nieuws/ondernemersnieuws/hoe-ervaren-mkbers-de-implementatie-van-ai-onderzoek-wijst-uit-dat-centrale) showed earlier this year that a central approach to AI in SMEs is largely absent. Only about 6% of entrepreneurs have actually integrated AI into daily operations. The rest use it here and there, ad hoc, often because one employee took out a ChatGPT account and is doing something useful with it. That is exactly where my father stood in March 1992, before we placed that advertisement. Seeing that something was happening, but not yet taking the step to actively go to market with it. I wrote earlier about the [exponential growth bias](/en/blog/surprised-from-my-ai-bubble) that causes entrepreneurs to structurally underestimate AI. Their planning is used to linear growth, while AI adoption develops along an exponential curve: first it feels slow, then suddenly very fast. What I want to add now is that recognising that curve is not enough. You also have to actually step on early, and that is precisely what the average SME director is not doing right now. ## What 1992 taught me that applies again in 2026 The directors I meet have good arguments for waiting. They want a business case first. They want to know what the ROI is. They want certainty about the EU AI Act and about privacy. They do not want to be the first, because pioneers get overtaken by faster followers. That is all true. And yet, in this case, waiting is the most expensive thing you can do. A 95-guilder advertisement in 1992 had no pre-calculated ROI. Neither did a website in 1997, and certainly not fully parting with the field force in 2008. All three were decisions we made on a gut feeling that we could afford a small gamble, and with a willingness to adjust if it did not work. With all three we could have been wrong. The difference with today's SME director is that AI in 2026 is already past the experimental stage. The technology works. The tools are here. The threshold to start is lower than ever, because you do not have to hire a developer to build your first agent. What is missing is not availability, it is the willingness to buy something whose return you cannot yet calculate. Waiting for the moment when AI has "crystallised" is the same as waiting on the internet until 2020. By then the competitor was already building a lead and a customer base that does not go away. ## What you can do today I am not a freelance consultant who is going to sell you a six-month programme, and this piece is not a run-up to that. But if I had to pass on one thing to an SME director who is still waiting: do something concrete, this month. Not much, but something. Concretely that means: pick one process with a lot of repetition and a clear owner, and put an AI tool on it that you can try out for less than a few hundred euros a month. Quote follow-up. First-line customer questions. Draft reports. Minutes. It matters less what it becomes, it matters that it is something and that someone in your organisation becomes responsible for it. The mistake you can then make is a few hundred euros a month, plus some lost hours. The mistake you make by doing nothing is a deficit that cannot be made up in two years' time because your competitor has by then redesigned the entire customer experience based on what they learned in those two years. My father faced the same choice in 1992, and again in 1997, and in 2008 that choice was on the table again for my sister and me. Three times the cost of trying was small, and three times the cost of not trying would have been much greater. That is not hindsight vision, it is calculating in advance with an honest estimate of what it costs to wait. ## In closing My father passed away in 2012. I was never able to ask him what he would have thought of AI, but I know that in 1992 he allowed an advertisement of 95 guilders to be placed that he was actually against. That is courage. In 2026 he would probably immediately see the potential and get to work with AI with as much enthusiasm as I do now. --- ## Sources - Sharp Netherlands & Wolters Kluwer European SME Study (December 2025): [Dutch SMEs leading Europe in AI ambitions](https://www.wolterskluwer.com/en/news/dutch-smes-are-leading-the-way-in-europe-in-terms-of-ai-ambitions-and-cloud-infrastructure) - Van Spaendonck Groep / MKB Servicedesk (March 2026): [AI not yet concretely integrated in SME businesses](https://www.mkbservicedesk.nl/nieuws/ondernemersnieuws/hoe-ervaren-mkbers-de-implementatie-van-ai-onderzoek-wijst-uit-dat-centrale) - Google: [Company history](https://about.google/company-info/our-history/) - Marc Diks: [Surprised from my AI bubble](/en/blog/surprised-from-my-ai-bubble) — earlier blog on exponential growth bias